wip

added python rag news summary

Dockerfile

@@ -5,8 +5,8 @@ ARG GOFLAGS="'-ldflags=-w -s'"
 
 WORKDIR /go/src/github.com/jmorganca/ollama
 RUN apt-get update && apt-get install -y git build-essential cmake
-ADD https://dl.google.com/go/go1.21.1.linux-$TARGETARCH.tar.gz /tmp/go1.21.1.tar.gz
-RUN mkdir -p /usr/local && tar xz -C /usr/local </tmp/go1.21.1.tar.gz
+ADD https://dl.google.com/go/go1.21.3.linux-$TARGETARCH.tar.gz /tmp/go1.21.3.tar.gz
+RUN mkdir -p /usr/local && tar xz -C /usr/local </tmp/go1.21.3.tar.gz
 
 COPY . .
 ENV GOARCH=$TARGETARCH

Dockerfile.build

@@ -1,6 +1,5 @@
-
 # centos7 amd64 dependencies
-FROM --platform=linux/amd64 nvidia/cuda:11.8.0-devel-centos7 AS base-amd64
+FROM --platform=linux/amd64 nvidia/cuda:11.3.1-devel-centos7 AS base-amd64
 RUN yum install -y https://repo.ius.io/ius-release-el7.rpm centos-release-scl && \
     yum update -y && \
     yum install -y devtoolset-10-gcc devtoolset-10-gcc-c++ git236 wget
@@ -8,7 +7,7 @@ RUN wget "https://github.com/Kitware/CMake/releases/download/v3.27.6/cmake-3.27.
 ENV PATH /opt/rh/devtoolset-10/root/usr/bin:$PATH
 
 # centos8 arm64 dependencies
-FROM --platform=linux/arm64 nvidia/cuda:11.4.3-devel-centos8 AS base-arm64
+FROM --platform=linux/arm64 nvidia/cuda-arm64:11.3.1-devel-centos8 AS base-arm64
 RUN sed -i -e 's/mirrorlist/#mirrorlist/g' -e 's|#baseurl=http://mirror.centos.org|baseurl=http://vault.centos.org|g' /etc/yum.repos.d/CentOS-*
 RUN yum install -y git cmake
 
@@ -17,8 +16,8 @@ ARG TARGETARCH
 ARG GOFLAGS="'-ldflags -w -s'"
 
 # install go
-ADD https://dl.google.com/go/go1.21.1.linux-$TARGETARCH.tar.gz /tmp/go1.21.1.tar.gz
-RUN mkdir -p /usr/local && tar xz -C /usr/local </tmp/go1.21.1.tar.gz
+ADD https://dl.google.com/go/go1.21.3.linux-$TARGETARCH.tar.gz /tmp/go1.21.3.tar.gz
+RUN mkdir -p /usr/local && tar xz -C /usr/local </tmp/go1.21.3.tar.gz
 
 # build the final binary
 WORKDIR /go/src/github.com/jmorganca/ollama

README.md

@@ -15,6 +15,10 @@ Get up and running with large language models locally.
 
 [Download](https://ollama.ai/download/Ollama-darwin.zip)
 
+### Windows
+
+Coming soon!
+
 ### Linux & WSL2
 
 ```
@@ -23,9 +27,9 @@ curl https://ollama.ai/install.sh | sh
 
 [Manual install instructions](https://github.com/jmorganca/ollama/blob/main/docs/linux.md)
 
-### Windows
+### Docker
 
-coming soon
+See the official [Docker image](https://hub.docker.com/r/ollama/ollama).
 
 ## Quickstart
 
@@ -56,11 +60,11 @@ Here are some example open-source models that can be downloaded:
 
 ## Customize your own model
 
-### Import from GGUF or GGML
+### Import from GGUF
 
-Ollama supports importing GGUF and GGML file formats in the Modelfile. This means if you have a model that is not in the Ollama library, you can create it, iterate on it, and upload it to the Ollama library to share with others when you are ready.
+Ollama supports importing GGUF models in the Modelfile:
 
-1. Create a file named Modelfile, and add a `FROM` instruction with the local filepath to the model you want to import.
+1. Create a file named `Modelfile`, with a `FROM` instruction with the local filepath to the model you want to import.
 
    ```
    FROM ./vicuna-33b.Q4_0.gguf
@@ -69,15 +73,19 @@ Ollama supports importing GGUF and GGML file formats in the Modelfile. This mean
 2. Create the model in Ollama
 
    ```
-   ollama create name -f path_to_modelfile
+   ollama create example -f Modelfile
    ```
 
 3. Run the model
 
    ```
-   ollama run name
+   ollama run example
    ```
 
+### Import from PyTorch or Safetensors
+
+See the [guide](docs/import.md) on importing models for more information.
+
 ### Customize a prompt
 
 Models from the Ollama library can be customized with a prompt. The example
@@ -195,7 +203,7 @@ Finally, in a separate shell, run a model:
 
 ## REST API
 
-> See the [API documentation](docs/api.md) for all endpoints.
+See the [API documentation](docs/api.md) for all endpoints.
 
 Ollama has an API for running and managing models. For example to generate text from a model:
 
@@ -222,3 +230,6 @@ curl -X POST http://localhost:11434/api/generate -d '{
 - [Typescript UI](https://github.com/ollama-interface/Ollama-Gui?tab=readme-ov-file)
 - [Dumbar](https://github.com/JerrySievert/Dumbar)
 - [Emacs client](https://github.com/zweifisch/ollama)
+- [oterm](https://github.com/ggozad/oterm)
+- [Ellama Emacs client](https://github.com/s-kostyaev/ellama)
+- [OllamaSharp for .NET](https://github.com/awaescher/OllamaSharp)

api/client.go

@@ -14,6 +14,7 @@ import (
 	"runtime"
 	"strings"
 
+	"github.com/jmorganca/ollama/format"
 	"github.com/jmorganca/ollama/version"
 )
 
@@ -51,8 +52,10 @@ func ClientFromEnvironment() (*Client, error) {
 	host, port, err := net.SplitHostPort(hostport)
 	if err != nil {
 		host, port = "127.0.0.1", "11434"
-		if ip := net.ParseIP(strings.Trim(os.Getenv("OLLAMA_HOST"), "[]")); ip != nil {
+		if ip := net.ParseIP(strings.Trim(hostport, "[]")); ip != nil {
 			host = ip.String()
+		} else if hostport != "" {
+			host = hostport
 		}
 	}
 
@@ -127,7 +130,7 @@ func (c *Client) do(ctx context.Context, method, path string, reqData, respData
 	return nil
 }
 
-const maxBufferSize = 512 * 1000 // 512KB
+const maxBufferSize = 512 * format.KiloByte
 
 func (c *Client) stream(ctx context.Context, method, path string, data any, fn func([]byte) error) error {
 	var buf *bytes.Buffer

api/types.go

@@ -3,7 +3,6 @@ package api
 import (
 	"encoding/json"
 	"fmt"
-	"log"
 	"math"
 	"os"
 	"reflect"
@@ -162,15 +161,10 @@ func (r *GenerateResponse) Summary() {
 	}
 }
 
-type Options struct {
-	Seed int `json:"seed,omitempty"`
-
-	// Backend options
-	UseNUMA bool `json:"numa,omitempty"`
-
-	// Model options
+// Runner options which must be set when the model is loaded into memory
+type Runner struct {
+	UseNUMA            bool    `json:"numa,omitempty"`
 	NumCtx             int     `json:"num_ctx,omitempty"`
-	NumKeep            int     `json:"num_keep,omitempty"`
 	NumBatch           int     `json:"num_batch,omitempty"`
 	NumGQA             int     `json:"num_gqa,omitempty"`
 	NumGPU             int     `json:"num_gpu,omitempty"`
@@ -184,8 +178,15 @@ type Options struct {
 	EmbeddingOnly      bool    `json:"embedding_only,omitempty"`
 	RopeFrequencyBase  float32 `json:"rope_frequency_base,omitempty"`
 	RopeFrequencyScale float32 `json:"rope_frequency_scale,omitempty"`
+	NumThread          int     `json:"num_thread,omitempty"`
+}
 
-	// Predict options
+type Options struct {
+	Runner
+
+	// Predict options used at runtime
+	NumKeep          int      `json:"num_keep,omitempty"`
+	Seed             int      `json:"seed,omitempty"`
 	NumPredict       int      `json:"num_predict,omitempty"`
 	TopK             int      `json:"top_k,omitempty"`
 	TopP             float32  `json:"top_p,omitempty"`
@@ -201,8 +202,6 @@ type Options struct {
 	MirostatEta      float32  `json:"mirostat_eta,omitempty"`
 	PenalizeNewline  bool     `json:"penalize_newline,omitempty"`
 	Stop             []string `json:"stop,omitempty"`
-
-	NumThread int `json:"num_thread,omitempty"`
 }
 
 var ErrInvalidOpts = fmt.Errorf("invalid options")
@@ -238,44 +237,39 @@ func (opts *Options) FromMap(m map[string]interface{}) error {
 						// when JSON unmarshals numbers, it uses float64, not int
 						field.SetInt(int64(t))
 					default:
-						log.Printf("could not convert model parameter %v of type %T to int, skipped", key, val)
+						return fmt.Errorf("option %q must be of type integer", key)
 					}
 				case reflect.Bool:
 					val, ok := val.(bool)
 					if !ok {
-						log.Printf("could not convert model parameter %v of type %T to bool, skipped", key, val)
-						continue
+						return fmt.Errorf("option %q must be of type boolean", key)
 					}
 					field.SetBool(val)
 				case reflect.Float32:
 					// JSON unmarshals to float64
 					val, ok := val.(float64)
 					if !ok {
-						log.Printf("could not convert model parameter %v of type %T to float32, skipped", key, val)
-						continue
+						return fmt.Errorf("option %q must be of type float32", key)
 					}
 					field.SetFloat(val)
 				case reflect.String:
 					val, ok := val.(string)
 					if !ok {
-						log.Printf("could not convert model parameter %v of type %T to string, skipped", key, val)
-						continue
+						return fmt.Errorf("option %q must be of type string", key)
 					}
 					field.SetString(val)
 				case reflect.Slice:
 					// JSON unmarshals to []interface{}, not []string
 					val, ok := val.([]interface{})
 					if !ok {
-						log.Printf("could not convert model parameter %v of type %T to slice, skipped", key, val)
-						continue
+						return fmt.Errorf("option %q must be of type array", key)
 					}
 					// convert []interface{} to []string
 					slice := make([]string, len(val))
 					for i, item := range val {
 						str, ok := item.(string)
 						if !ok {
-							log.Printf("could not convert model parameter %v of type %T to slice of strings, skipped", key, item)
-							continue
+							return fmt.Errorf("option %q must be of an array of strings", key)
 						}
 						slice[i] = str
 					}
@@ -315,20 +309,22 @@ func DefaultOptions() Options {
 		PenalizeNewline:  true,
 		Seed:             -1,
 
-		// options set when the model is loaded
-		NumCtx:             2048,
-		RopeFrequencyBase:  10000.0,
-		RopeFrequencyScale: 1.0,
-		NumBatch:           512,
-		NumGPU:             -1, // -1 here indicates that NumGPU should be set dynamically
-		NumGQA:             1,
-		NumThread:          0, // let the runtime decide
-		LowVRAM:            false,
-		F16KV:              true,
-		UseMLock:           false,
-		UseMMap:            true,
-		UseNUMA:            false,
-		EmbeddingOnly:      true,
+		Runner: Runner{
+			// options set when the model is loaded
+			NumCtx:             2048,
+			RopeFrequencyBase:  10000.0,
+			RopeFrequencyScale: 1.0,
+			NumBatch:           512,
+			NumGPU:             -1, // -1 here indicates that NumGPU should be set dynamically
+			NumGQA:             1,
+			NumThread:          0, // let the runtime decide
+			LowVRAM:            false,
+			F16KV:              true,
+			UseMLock:           false,
+			UseMMap:            true,
+			UseNUMA:            false,
+			EmbeddingOnly:      true,
+		},
 	}
 }
 

app/forge.config.ts

@@ -47,16 +47,6 @@ const config: ForgeConfig = {
   },
   rebuildConfig: {},
   makers: [new MakerSquirrel({}), new MakerZIP({}, ['darwin'])],
-  publishers: [
-    new PublisherGithub({
-      repository: {
-        name: 'ollama',
-        owner: 'jmorganca',
-      },
-      draft: false,
-      prerelease: true,
-    }),
-  ],
   hooks: {
     readPackageJson: async (_, packageJson) => {
       return { ...packageJson, version: process.env.VERSION || packageJson.version }

app/package.json

@@ -46,7 +46,7 @@
     "chmodr": "^1.2.0",
     "copy-webpack-plugin": "^11.0.0",
     "css-loader": "^6.8.1",
-    "electron": "25.2.0",
+    "electron": "25.9.2",
     "eslint": "^8.43.0",
     "eslint-plugin-import": "^2.27.5",
     "fork-ts-checker-webpack-plugin": "^7.3.0",

app/src/index.ts

@@ -162,13 +162,56 @@ app.on('before-quit', () => {
   }
 })
 
+const updateURL = `https://ollama.ai/api/update?os=${process.platform}&arch=${
+  process.arch
+}&version=${app.getVersion()}&id=${id()}`
+
+let latest = ''
+async function isNewReleaseAvailable() {
+  try {
+    const response = await fetch(updateURL)
+
+    if (!response.ok) {
+      return false
+    }
+
+    if (response.status === 204) {
+      return false
+    }
+
+    const data = await response.json()
+
+    const url = data?.url
+    if (!url) {
+      return false
+    }
+
+    if (latest === url) {
+      return false
+    }
+
+    latest = url
+
+    return true
+  } catch (error) {
+    logger.error(`update check failed - ${error}`)
+    return false
+  }
+}
+
+async function checkUpdate() {
+  const available = await isNewReleaseAvailable()
+  if (available) {
+    logger.info('checking for update')
+    autoUpdater.checkForUpdates()
+  }
+}
+
 function init() {
   if (app.isPackaged) {
-    autoUpdater.checkForUpdates()
+    checkUpdate()
     setInterval(() => {
-      if (!updateAvailable) {
-        autoUpdater.checkForUpdates()
-      }
+      checkUpdate()
     }, 60 * 60 * 1000)
   }
 
@@ -246,11 +289,7 @@ function id(): string {
   return uuid
 }
 
-autoUpdater.setFeedURL({
-  url: `https://ollama.ai/api/update?os=${process.platform}&arch=${
-    process.arch
-  }&version=${app.getVersion()}&id=${id()}`,
-})
+autoUpdater.setFeedURL({ url: updateURL })
 
 autoUpdater.on('error', e => {
   logger.error(`update check failed - ${e.message}`)

cmd/cmd.go

@@ -78,18 +78,12 @@ func CreateHandler(cmd *cobra.Command, args []string) error {
 				spinner.Stop()
 			}
 			currentDigest = resp.Digest
-			switch {
-			case strings.Contains(resp.Status, "embeddings"):
-				bar = progressbar.Default(resp.Total, resp.Status)
-				bar.Set64(resp.Completed)
-			default:
-				// pulling
-				bar = progressbar.DefaultBytes(
-					resp.Total,
-					resp.Status,
-				)
-				bar.Set64(resp.Completed)
-			}
+			// pulling
+			bar = progressbar.DefaultBytes(
+				resp.Total,
+				resp.Status,
+			)
+			bar.Set64(resp.Completed)
 		} else if resp.Digest == currentDigest && resp.Digest != "" {
 			bar.Set64(resp.Completed)
 		} else {
@@ -694,7 +688,12 @@ func generateInteractive(cmd *cobra.Command, model string) error {
 		case strings.HasPrefix(line, "/show"):
 			args := strings.Fields(line)
 			if len(args) > 1 {
-				resp, err := server.GetModelInfo(model)
+				client, err := api.ClientFromEnvironment()
+				if err != nil {
+					fmt.Println("error: couldn't connect to ollama server")
+					return err
+				}
+				resp, err := client.Show(cmd.Context(), &api.ShowRequest{Name: model})
 				if err != nil {
 					fmt.Println("error: couldn't get model")
 					return err
@@ -933,7 +932,7 @@ func NewCLI() *cobra.Command {
 	createCmd := &cobra.Command{
 		Use:     "create MODEL",
 		Short:   "Create a model from a Modelfile",
-		Args:    cobra.MinimumNArgs(1),
+		Args:    cobra.ExactArgs(1),
 		PreRunE: checkServerHeartbeat,
 		RunE:    CreateHandler,
 	}
@@ -943,7 +942,7 @@ func NewCLI() *cobra.Command {
 	showCmd := &cobra.Command{
 		Use:     "show MODEL",
 		Short:   "Show information for a model",
-		Args:    cobra.MinimumNArgs(1),
+		Args:    cobra.ExactArgs(1),
 		PreRunE: checkServerHeartbeat,
 		RunE:    ShowHandler,
 	}
@@ -970,13 +969,14 @@ func NewCLI() *cobra.Command {
 		Use:     "serve",
 		Aliases: []string{"start"},
 		Short:   "Start ollama",
+		Args:    cobra.ExactArgs(0),
 		RunE:    RunServer,
 	}
 
 	pullCmd := &cobra.Command{
 		Use:     "pull MODEL",
 		Short:   "Pull a model from a registry",
-		Args:    cobra.MinimumNArgs(1),
+		Args:    cobra.ExactArgs(1),
 		PreRunE: checkServerHeartbeat,
 		RunE:    PullHandler,
 	}
@@ -986,7 +986,7 @@ func NewCLI() *cobra.Command {
 	pushCmd := &cobra.Command{
 		Use:     "push MODEL",
 		Short:   "Push a model to a registry",
-		Args:    cobra.MinimumNArgs(1),
+		Args:    cobra.ExactArgs(1),
 		PreRunE: checkServerHeartbeat,
 		RunE:    PushHandler,
 	}
@@ -1002,15 +1002,15 @@ func NewCLI() *cobra.Command {
 	}
 
 	copyCmd := &cobra.Command{
-		Use:     "cp",
+		Use:     "cp SOURCE TARGET",
 		Short:   "Copy a model",
-		Args:    cobra.MinimumNArgs(2),
+		Args:    cobra.ExactArgs(2),
 		PreRunE: checkServerHeartbeat,
 		RunE:    CopyHandler,
 	}
 
 	deleteCmd := &cobra.Command{
-		Use:     "rm",
+		Use:     "rm MODEL [MODEL...]",
 		Short:   "Remove a model",
 		Args:    cobra.MinimumNArgs(1),
 		PreRunE: checkServerHeartbeat,

docs/api.md

@@ -355,7 +355,7 @@ curl -X POST http://localhost:11434/api/embeddings -d '{
 
 ```json
 {
-  "embeddings": [
+  "embedding": [
     0.5670403838157654, 0.009260174818336964, 0.23178744316101074, -0.2916173040866852, -0.8924556970596313,
     0.8785552978515625, -0.34576427936553955, 0.5742510557174683, -0.04222835972905159, -0.137906014919281
   ]

docs/faq.md

@@ -1,5 +1,21 @@
 # FAQ
 
+## How can I view the logs?
+
+On macOS:
+
+```
+cat ~/.ollama/logs/server.log
+```
+
+On Linux:
+
+```
+journalctl -u ollama
+```
+
+If you're running `ollama serve` directly, the logs will be printed to the console.
+
 ## How can I expose the Ollama server?
 
 ```bash
@@ -14,5 +30,5 @@ OLLAMA_ORIGINS=http://192.168.1.1:*,https://example.com ollama serve
 
 ## Where are models stored?
 
-* macOS: Raw model data is stored under `~/.ollama/models`.
-* Linux: Raw model data is stored under `/usr/share/ollama/.ollama/models`
+- macOS: Raw model data is stored under `~/.ollama/models`.
+- Linux: Raw model data is stored under `/usr/share/ollama/.ollama/models`

docs/import.md

@@ -0,0 +1,163 @@
+# Import a model
+
+This guide walks through importing a PyTorch, Safetensors or GGUF model.
+
+## Supported models
+
+Ollama supports a set of model architectures, with support for more coming soon:
+
+- Llama & Mistral
+- Falcon & RW
+- GPT-NeoX
+- BigCode
+
+To view a model's architecture, check the `config.json` file in its HuggingFace repo. You should see an entry under `architectures` (e.g. `LlamaForCausalLM`).
+
+## Importing
+
+### Step 1: Clone the HuggingFace repository (optional)
+
+If the model is currently hosted in a HuggingFace repository, first clone that repository to download the raw model.
+
+```
+git lfs install
+git clone https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.1
+cd Mistral-7B-Instruct-v0.1
+```
+
+### Step 2: Convert and quantize to a `.bin` file (optional, for PyTorch and Safetensors)
+
+If the model is in PyTorch or Safetensors format, a [Docker image](https://hub.docker.com/r/ollama/quantize) with the tooling required to convert and quantize models is available.
+
+First, Install [Docker](https://www.docker.com/get-started/).
+
+Next, to convert and quantize your model, run:
+
+```
+docker run --rm -v .:/model ollama/quantize -q q4_0 /model
+```
+
+This will output two files into the directory:
+
+- `f16.bin`: the model converted to GGUF
+- `q4_0.bin` the model quantized to a 4-bit quantization (we will use this file to create the Ollama model)
+
+### Step 3: Write a `Modelfile`
+
+Next, create a `Modelfile` for your model. This file is the blueprint for your model, specifying weights, parameters, prompt templates and more.
+
+```
+FROM ./q4_0.bin
+```
+
+(Optional) many chat models require a prompt template in order to answer correctly. A default prompt template can be specified with the `TEMPLATE` instruction in the `Modelfile`:
+
+```
+FROM ./q4_0.bin
+TEMPLATE "[INST] {{ .Prompt }} [/INST]"
+```
+
+### Step 4: Create the Ollama model
+
+Finally, create a model from your `Modelfile`:
+
+```
+ollama create example -f Modelfile
+```
+
+Next, test the model with `ollama run`:
+
+```
+ollama run example "What is your favourite condiment?"
+```
+
+### Step 5: Publish your model (optional – early alpha)
+
+Publishing models is in early alpha. If you'd like to publish your model to share with others, follow these steps:
+
+1. Create [an account](https://ollama.ai/signup)
+2. Run `cat ~/.ollama/id_ed25519.pub` to view your Ollama public key. Copy this to the clipboard.
+3. Add your public key to your [Ollama account](https://ollama.ai/settings/keys)
+
+Next, copy your model to your username's namespace:
+
+```
+ollama cp example <your username>/example
+```
+
+Then push the model:
+
+```
+ollama push <your username>/example
+```
+
+After publishing, your model will be available at `https://ollama.ai/<your username>/example`.
+
+## Quantization reference
+
+The quantization options are as follow (from highest highest to lowest levels of quantization). Note: some architectures such as Falcon do not support K quants.
+
+- `q2_K`
+- `q3_K`
+- `q3_K_S`
+- `q3_K_M`
+- `q3_K_L`
+- `q4_0` (recommended)
+- `q4_1`
+- `q4_K`
+- `q4_K_S`
+- `q4_K_M`
+- `q5_0`
+- `q5_1`
+- `q5_K`
+- `q5_K_S`
+- `q5_K_M`
+- `q6_K`
+- `q8_0`
+
+## Manually converting & quantizing models
+
+### Prerequisites
+
+Start by cloning the `llama.cpp` repo to your machine in another directory:
+
+```
+git clone https://github.com/ggerganov/llama.cpp.git
+cd llama.cpp
+```
+
+Next, install the Python dependencies:
+
+```
+pip install -r requirements.txt
+```
+
+Finally, build the `quantize` tool:
+
+```
+make quantize
+```
+
+### Convert the model
+
+Run the correct conversion script for your model architecture:
+
+```shell
+# LlamaForCausalLM or MistralForCausalLM
+python convert.py <path to model directory>
+
+# FalconForCausalLM
+python convert-falcon-hf-to-gguf.py <path to model directory>
+
+# GPTNeoXForCausalLM
+python convert-falcon-hf-to-gguf.py <path to model directory>
+
+# GPTBigCodeForCausalLM
+python convert-starcoder-hf-to-gguf.py <path to model directory>
+```
+
+### Quantize the model
+
+```
+quantize <path to model dir>/ggml-model-f32.bin <path to model dir>/q4_0.bin q4_0
+```

docs/linux.md

@@ -80,4 +80,3 @@ To view logs of Ollama running as a startup service, run:
 ```bash
 journalctl -u ollama
 ```
-

docs/modelfile.md

@@ -12,7 +12,6 @@ A model file is the blueprint to create and share models with Ollama.
   - [FROM (Required)](#from-required)
     - [Build from llama2](#build-from-llama2)
     - [Build from a bin file](#build-from-a-bin-file)
-  - [EMBED](#embed)
   - [PARAMETER](#parameter)
     - [Valid Parameters and Values](#valid-parameters-and-values)
   - [TEMPLATE](#template)
@@ -91,17 +90,6 @@ FROM ./ollama-model.bin
 
 This bin file location should be specified as an absolute path or relative to the `Modelfile` location.
 
-### EMBED
-
-The `EMBED` instruction is used to add embeddings of files to a model. This is useful for adding custom data that the model can reference when generating an answer. Note that currently only text files are supported, formatted with each line as one embedding.
-
-```modelfile
-FROM <model name>:<tag>
-EMBED <file path>.txt
-EMBED <different file path>.txt
-EMBED <path to directory>/*.txt
-```
-
 ### PARAMETER
 
 The `PARAMETER` instruction defines a parameter that can be set when the model is run.
@@ -124,7 +112,7 @@ PARAMETER <parameter> <parametervalue>
 | repeat_last_n  | Sets how far back for the model to look back to prevent repetition. (Default: 64, 0 = disabled, -1 = num_ctx)                                                                                                                                           | int        | repeat_last_n 64     |
 | repeat_penalty | Sets how strongly to penalize repetitions. A higher value (e.g., 1.5) will penalize repetitions more strongly, while a lower value (e.g., 0.9) will be more lenient. (Default: 1.1)                                                                     | float      | repeat_penalty 1.1   |
 | temperature    | The temperature of the model. Increasing the temperature will make the model answer more creatively. (Default: 0.8)                                                                                                                                     | float      | temperature 0.7      |
-| seed | Sets the random number seed to use for generation. Setting this to a specific number will make the model generate the same text for the same prompt. | int | seed 42 |
+| seed | Sets the random number seed to use for generation. Setting this to a specific number will make the model generate the same text for the same prompt. (Default: 0) | int | seed 42 |
 | stop           | Sets the stop sequences to use.                                                                                                                                                                                                                         | string     | stop "AI assistant:" |
 | tfs_z          | Tail free sampling is used to reduce the impact of less probable tokens from the output. A higher value (e.g., 2.0) will reduce the impact more, while a value of 1.0 disables this setting. (default: 1)                                               | float      | tfs_z 1              |
 | num_predict    | Maximum number of tokens to predict when generating text. (Default: 128, -1 = infinite generation, -2 = fill context)                                                                                                                                   | int        | num_predict 42       |

docs/quantize.md

@@ -1,111 +0,0 @@
-# How to Quantize a Model
-
-Sometimes the model you want to work with is not available at [https://ollama.ai/library](https://ollama.ai/library).
-
-## Figure out if we can run the model?
-
-Not all models will work with Ollama. There are a number of factors that go into whether we are able to work with the next cool model. First it has to work with llama.cpp. Then we have to have implemented the features of llama.cpp that it requires. And then, sometimes, even with both of those, the model might not work...
-
-1. What is the model you want to convert and upload?
-2. Visit the model's page on HuggingFace.
-3. Switch to the **Files and versions** tab.
-4. Click on the **config.json** file. If there is no config.json file, it may not work.
-5. Take note of the **architecture** list in the json file.
-6. Does any entry in the list match one of the following architectures?
-    1. LlamaForCausalLM
-    2. MistralForCausalLM
-    3. RWForCausalLM
-    4. FalconForCausalLM
-    5. GPTNeoXForCausalLM
-    6. GPTBigCodeForCausalLM
-7. If the answer is yes, then there is a good chance the model will run after being converted and quantized.
-8. An alternative to this process is to visit [https://caniquant.tvl.st](https://caniquant.tvl.st) and enter the org/modelname in the box and submit.
-
-At this point there are two processes you can use. You can either use a Docker container to convert and quantize, OR you can manually run the scripts. The Docker container is the easiest way to do it, but it requires you to have Docker installed on your machine. If you don't have Docker installed, you can follow the manual process.
-
-## Convert and Quantize with Docker
-
-Run `docker run --rm -v /path/to/model/repo:/repo ollama/quantize -q quantlevel /repo`. For instance, if you have downloaded the latest Mistral 7B model, then clone it to your machine. Then change into that directory and you can run:
-
-```shell
-docker run --rm -v .:/repo ollama/quantize -q q4_0 /repo
-```
-
-You can find the different quantization levels below under **Quantize the Model**.
-
-This will output two files into the directory. First is a f16.bin file that is the model converted to GGUF. The second file is a q4_0.bin file which is the model quantized to a 4 bit quantization. You should rename it to something more descriptive.
-
-You can find the repository for the Docker container here: [https://github.com/mxyng/quantize](https://github.com/mxyng/quantize)
-
-For instance, if you wanted to convert the Mistral 7B model to a Q4 quantized model, then you could go through the following steps:
-
-1. First verify the model will potentially work.
-2. Now clone Mistral 7B to your machine. You can find the command to run when you click the three vertical dots button on the model page, then click **Clone Repository**.
-   1. For this repo, the command is:
-
-      ```shell
-      git lfs install
-      git clone https://huggingface.co/mistralai/Mistral-7B-v0.1
-      ```
-
-   2. Navigate into the new directory and run `docker run --rm -v .:/repo ollama/quantize -q q4_0 /repo`
-   3. Now you can create a modelfile using the q4_0.bin file that was created.
-
-## Convert and Quantize Manually
-
-### Clone llama.cpp to your machine
-
-If we know the model has a chance of working, then we need to convert and quantize. This is a matter of running two separate scripts in the llama.cpp project.
-
-1. Decide where you want the llama.cpp repository on your machine.
-2. Navigate to that location and then run:
- [`git clone https://github.com/ggerganov/llama.cpp.git`](https://github.com/ggerganov/llama.cpp.git)
-    1. If you don't have git installed, download this zip file and unzip it to that location: https://github.com/ggerganov/llama.cpp/archive/refs/heads/master.zip
-3. Install the Python dependencies: `pip install torch transformers sentencepiece`
-4. Run 'make' to build the project and the quantize executable.
-
-### Convert the model to GGUF
-
-1. Decide on the right convert script to run. What was the model architecture you found in the first section.
-    1. LlamaForCausalLM or MistralForCausalLM:
-    run `python3 convert.py <modelfilename>`
-    No need to specify fp16 or fp32.
-    2. FalconForCausalLM or RWForCausalLM:
-    run `python3 convert-falcon-hf-to-gguf.py <modelfilename> <fpsize>`  
-    fpsize depends on the weight size. 1 for fp16, 0 for fp32
-    3. GPTNeoXForCausalLM:
-    run `python3 convert-gptneox-hf-to-gguf.py <modelfilename> <fpsize>`
-    fpsize depends on the weight size. 1 for fp16, 0 for fp32
-    4. GPTBigCodeForCausalLM:
-    run `python3 convert-starcoder-hf-to-gguf.py <modelfilename> <fpsize>`
-    fpsize depends on the weight size. 1 for fp16, 0 for fp32
-
-### Quantize the model
-
-If the model converted successfully, there is a good chance it will also quantize successfully. Now you need to decide on the quantization to use. We will always try to create all the quantizations and upload them to the library. You should decide which level is more important to you and quantize accordingly.
-
-The quantization options are as follows. Note that some architectures such as Falcon do not support K quants.
-
-- Q4_0
-- Q4_1
-- Q5_0
-- Q5_1
-- Q2_K
-- Q3_K
-- Q3_K_S
-- Q3_K_M
-- Q3_K_L
-- Q4_K
-- Q4_K_S
-- Q4_K_M
-- Q5_K
-- Q5_K_S
-- Q5_K_M
-- Q6_K
-- Q8_0
-
-Run the following command `quantize <converted model from above> <output file> <quantization type>`
-
-## Now Create the Model
-
-Now you can create the Ollama model. Refer to the [modelfile](./modelfile.md) doc for more information on doing that.

examples/golang-simplegenerate/main.go

@@ -3,10 +3,10 @@ package main
 import (
 	"bytes"
 	"fmt"
-	"net/http"
-	"os"
 	"io"
 	"log"
+	"net/http"
+	"os"
 )
 
 func main() {
@@ -16,7 +16,7 @@ func main() {
 	if err != nil {
 		fmt.Print(err.Error())
 		os.Exit(1)
-	} 
+	}
 
 	responseData, err := io.ReadAll(resp.Body)
 	if err != nil {

examples/python-rag-newssummary/README.md

@@ -0,0 +1,22 @@
+# News Summarizer
+
+This example goes through a series of steps:
+
+  1. You choose a topic area (e.g., "news", "NVidia", "music", etc.).
+  2. Gets the most recent articles on that topic from various sources.
+  3. Uses Ollama to summarize each article.
+  4. Creates chunks of sentences from each article.
+  5. Uses Sentence Transformers to generate embeddings for each of those chunks.
+  6. You enter a question regarding the summaries shown.
+  7. Uses Sentence Transformers to generate an embedding for that question.
+  8. Uses the embedded question to find the most similar chunks.
+  9. Feeds all that to Ollama to generate a good answer to your question based on these news articles.
+
+This example lets you pick from a few different topic areas, then summarize the most recent x articles for that topic. It then creates chunks of sentences from each article and then generates embeddings for each of those chunks.
+
+You can run the example like this:
+
+```bash
+pip install -r requirements.txt
+python summ.py
+```

examples/python-rag-newssummary/requirements.txt

@@ -0,0 +1,9 @@
+beautifulsoup4==4.12.2
+feedparser==6.0.10
+mattsollamatools==0.0.8
+newspaper3k==0.2.8
+nltk==3.8.1
+numpy==1.24.3
+Requests==2.31.0
+scikit_learn==1.3.0
+sentence_transformers==2.2.2

examples/python-rag-newssummary/summ.py

@@ -0,0 +1,86 @@
+import curses
+import json
+from utils import get_url_for_topic, topic_urls, menu, getUrls, get_summary, getArticleText, knn_search
+import requests
+from sentence_transformers import SentenceTransformer
+from mattsollamatools import chunker
+
+if __name__ == "__main__":
+    chosen_topic = curses.wrapper(menu)
+    print("Here is your news summary:\n")
+    urls = getUrls(chosen_topic, n=5)
+    model = SentenceTransformer('all-MiniLM-L6-v2')
+    allEmbeddings = []
+
+    for url in urls:
+      article={}
+      article['embeddings'] = []
+      article['url'] = url
+      text = getArticleText(url)
+      summary = get_summary(text)
+      chunks = chunker(text)  # Use the chunk_text function from web_utils
+      embeddings = model.encode(chunks)
+      for (chunk, embedding) in zip(chunks, embeddings):
+        item = {}
+        item['source'] = chunk
+        item['embedding'] = embedding.tolist()  # Convert NumPy array to list
+        item['sourcelength'] = len(chunk)
+        article['embeddings'].append(item)
+    
+      allEmbeddings.append(article)
+
+      print(f"{summary}\n")
+
+    
+    while True:
+      context = []
+      # Input a question from the user
+      question = input("Enter your question about the news, or type quit: ")
+
+      if question.lower() == 'quit':
+        break
+
+      # Embed the user's question
+      question_embedding = model.encode([question])
+
+      # Perform KNN search to find the best matches (indices and source text)
+      best_matches = knn_search(question_embedding, allEmbeddings, k=10)
+
+
+      sourcetext=""
+      for i, (index, source_text) in enumerate(best_matches, start=1):
+          sourcetext += f"{i}. Index: {index}, Source Text: {source_text}"
+
+      systemPrompt = f"Only use the following information to answer the question. Do not use anything else: {sourcetext}"
+
+      url = "http://localhost:11434/api/generate"
+
+      payload = {
+      "model": "mistral-openorca",
+      "prompt": question, 
+      "system": systemPrompt,
+      "stream": False, 
+      "context": context
+      }
+
+      # Convert the payload to a JSON string
+      payload_json = json.dumps(payload)
+
+      # Set the headers to specify JSON content
+      headers = {
+          "Content-Type": "application/json"
+      }
+
+      # Send the POST request
+      response = requests.post(url, data=payload_json, headers=headers)
+
+      # Check the response
+      if response.status_code == 200:
+          output = json.loads(response.text)
+          context = output['context']
+          print(output['response']+ "\n")
+          
+
+      else:
+          print(f"Request failed with status code {response.status_code}")
+

examples/python-rag-newssummary/utils.py

@@ -0,0 +1,108 @@
+import curses
+import feedparser
+import requests
+import unicodedata
+import json
+from newspaper import Article
+from bs4 import BeautifulSoup
+from nltk.tokenize import sent_tokenize, word_tokenize
+import numpy as np
+from sklearn.neighbors import NearestNeighbors
+from mattsollamatools import chunker
+
+# Create a dictionary to store topics and their URLs
+topic_urls = {
+    "Mac": "https://9to5mac.com/guides/mac/feed",
+    "News": "http://www.npr.org/rss/rss.php?id=1001",
+    "Nvidia": "https://nvidianews.nvidia.com/releases.xml",
+    "Raspberry Pi": "https://www.raspberrypi.com/news/feed/", 
+    "Music": "https://www.billboard.com/c/music/music-news/feed/"
+}
+
+# Use curses to create a menu of topics
+def menu(stdscr):
+    chosen_topic = get_url_for_topic(stdscr)  
+    url = topic_urls[chosen_topic] if chosen_topic in topic_urls else "Topic not found"
+    
+    stdscr.addstr(len(topic_urls) + 3, 0, f"Selected URL for {chosen_topic}: {url}")
+    stdscr.refresh()
+    
+    return chosen_topic
+
+# You have chosen a topic. Now return the url for that topic
+def get_url_for_topic(stdscr):
+    curses.curs_set(0)  # Hide the cursor
+    stdscr.clear()
+
+    stdscr.addstr(0, 0, "Choose a topic using the arrow keys (Press Enter to select):")
+
+    # Create a list of topics
+    topics = list(topic_urls.keys())
+    current_topic = 0
+
+    while True:
+        for i, topic in enumerate(topics):
+            if i == current_topic:
+                stdscr.addstr(i + 2, 2, f"> {topic}")
+            else:
+                stdscr.addstr(i + 2, 2, f"  {topic}")
+
+        stdscr.refresh()
+
+        key = stdscr.getch()
+
+        if key == curses.KEY_DOWN and current_topic < len(topics) - 1:
+            current_topic += 1
+        elif key == curses.KEY_UP and current_topic > 0:
+            current_topic -= 1
+        elif key == 10:  # Enter key
+            return topic_urls[topics[current_topic]]
+
+# Get the last N URLs from an RSS feed
+def getUrls(feed_url, n=20):
+    feed = feedparser.parse(feed_url)
+    entries = feed.entries[-n:]
+    urls = [entry.link for entry in entries]
+    return urls
+
+# Often there are a bunch of ads and menus on pages for a news article. This uses newspaper3k to get just the text of just the article.
+def getArticleText(url):
+  article = Article(url)
+  article.download()
+  article.parse()
+  return article.text
+
+def get_summary(text):
+  systemPrompt = "Write a concise summary of the text, return your responses with 5 lines that cover the key points of the text given."
+  prompt = text
+  
+  url = "http://localhost:11434/api/generate"
+
+  payload = {
+    "model": "mistral-openorca",
+    "prompt": prompt, 
+    "system": systemPrompt,
+    "stream": False
+  }
+  payload_json = json.dumps(payload)
+  headers = {"Content-Type": "application/json"}
+  response = requests.post(url, data=payload_json, headers=headers)
+
+  return json.loads(response.text)["response"]
+
+# Perform K-nearest neighbors (KNN) search
+def knn_search(question_embedding, embeddings, k=5):
+    X = np.array([item['embedding'] for article in embeddings for item in article['embeddings']])
+    source_texts = [item['source'] for article in embeddings for item in article['embeddings']]
+    
+    # Fit a KNN model on the embeddings
+    knn = NearestNeighbors(n_neighbors=k, metric='cosine')
+    knn.fit(X)
+    
+    # Find the indices and distances of the k-nearest neighbors
+    distances, indices = knn.kneighbors(question_embedding, n_neighbors=k)
+    
+    # Get the indices and source texts of the best matches
+    best_matches = [(indices[0][i], source_texts[indices[0][i]]) for i in range(k)]
+    
+    return best_matches

format/bytes.go

@@ -2,14 +2,21 @@ package format
 
 import "fmt"
 
+const (
+	Byte     = 1
+	KiloByte = Byte * 1000
+	MegaByte = KiloByte * 1000
+	GigaByte = MegaByte * 1000
+)
+
 func HumanBytes(b int64) string {
 	switch {
-	case b > 1000*1000*1000:
-		return fmt.Sprintf("%d GB", b/1000/1000/1000)
-	case b > 1000*1000:
-		return fmt.Sprintf("%d MB", b/1000/1000)
-	case b > 1000:
-		return fmt.Sprintf("%d KB", b/1000)
+	case b > GigaByte:
+		return fmt.Sprintf("%d GB", b/GigaByte)
+	case b > MegaByte:
+		return fmt.Sprintf("%d MB", b/MegaByte)
+	case b > KiloByte:
+		return fmt.Sprintf("%d KB", b/KiloByte)
 	default:
 		return fmt.Sprintf("%d B", b)
 	}

format/time_test.go

@@ -29,7 +29,7 @@ func TestHumanTime(t *testing.T) {
 	})
 
 	t.Run("soon", func(t *testing.T) {
-		v := now.Add(800*time.Millisecond)
+		v := now.Add(800 * time.Millisecond)
 		assertEqual(t, HumanTime(v, ""), "Less than a second from now")
 	})
 }

go.mod

@@ -45,7 +45,6 @@ require (
 	golang.org/x/sys v0.11.0 // indirect
 	golang.org/x/term v0.10.0
 	golang.org/x/text v0.10.0 // indirect
-	gonum.org/v1/gonum v0.13.0
 	google.golang.org/protobuf v1.30.0 // indirect
 	gopkg.in/yaml.v3 v3.0.1 // indirect
 )

go.sum

@@ -145,8 +145,6 @@ golang.org/x/text v0.10.0 h1:UpjohKhiEgNc0CSauXmwYftY1+LlaC75SJwh0SgCX58=
 golang.org/x/text v0.10.0/go.mod h1:TvPlkZtksWOMsz7fbANvkp4WM8x/WCo/om8BMLbz+aE=
 golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
 golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
-gonum.org/v1/gonum v0.13.0 h1:a0T3bh+7fhRyqeNbiC3qVHYmkiQgit3wnNan/2c0HMM=
-gonum.org/v1/gonum v0.13.0/go.mod h1:/WPYRckkfWrhWefxyYTfrTtQR0KH4iyHNuzxqXAKyAU=
 google.golang.org/protobuf v1.26.0-rc.1/go.mod h1:jlhhOSvTdKEhbULTjvd4ARK9grFBp09yW+WbY/TyQbw=
 google.golang.org/protobuf v1.28.0/go.mod h1:HV8QOd/L58Z+nl8r43ehVNZIU/HEI6OcFqwMG9pJV4I=
 google.golang.org/protobuf v1.30.0 h1:kPPoIgf3TsEvrm0PFe15JQ+570QVxYzEvvHqChK+cng=

llm/llama.cpp/patches/0001-remove-warm-up-logging.patch

@@ -1,6 +1,6 @@
-From 07993bdc35345b67b27aa649a7c099ad42d80c4c Mon Sep 17 00:00:00 2001
-From: Michael Yang <mxyng@pm.me>
-Date: Thu, 21 Sep 2023 14:43:21 -0700
+From 8dbb5449db259a9c24796e7927d89bee98b6c8f5 Mon Sep 17 00:00:00 2001
+From: Bruce MacDonald <brucewmacdonald@gmail.com>
+Date: Thu, 5 Oct 2023 11:21:12 -0400
 Subject: [PATCH] remove warm up logging
 
 ---
@@ -8,18 +8,18 @@ Subject: [PATCH] remove warm up logging
  1 file changed, 2 deletions(-)
 
 diff --git a/common/common.cpp b/common/common.cpp
-index 2597ba0..b56549b 100644
+index 7370017..c4433fe 100644
 --- a/common/common.cpp
 +++ b/common/common.cpp
-@@ -780,8 +780,6 @@ std::tuple<struct llama_model *, struct llama_context *> llama_init_from_gpt_par
+@@ -839,8 +839,6 @@ std::tuple<struct llama_model *, struct llama_context *> llama_init_from_gpt_par
      }
  
      {
 -        LOG("warming up the model with an empty run\n");
 -
-         const std::vector<llama_token> tmp = { llama_token_bos(lctx), llama_token_eos(lctx), };
-         llama_eval(lctx, tmp.data(), std::min(tmp.size(), (size_t) params.n_batch), 0, params.n_threads);
-         llama_reset_timings(lctx);
+         std::vector<llama_token> tmp = { llama_token_bos(lctx), llama_token_eos(lctx), };
+         llama_decode(lctx, llama_batch_get_one(tmp.data(), std::min(tmp.size(), (size_t) params.n_batch), 0, 0));
+         llama_kv_cache_tokens_rm(lctx, -1, -1);
 -- 
-2.42.0
+2.39.2 (Apple Git-143)
 

llm/llama.go

@@ -24,51 +24,53 @@ import (
 	"time"
 
 	"github.com/jmorganca/ollama/api"
+	"github.com/jmorganca/ollama/format"
 )
 
 //go:embed llama.cpp/*/build/*/bin/*
 var llamaCppEmbed embed.FS
 
 type ModelRunner struct {
-	Path string // path to the model runner executable
+	Path        string // path to the model runner executable
+	Accelerated bool
 }
 
 func chooseRunners(workDir, runnerType string) []ModelRunner {
 	buildPath := path.Join("llama.cpp", runnerType, "build")
-	var runners []string
+	var runners []ModelRunner
 
 	// set the runners based on the OS
 	// IMPORTANT: the order of the runners in the array is the priority order
 	switch runtime.GOOS {
 	case "darwin":
-		runners = []string{
-			path.Join(buildPath, "metal", "bin", "ollama-runner"),
-			path.Join(buildPath, "cpu", "bin", "ollama-runner"),
+		runners = []ModelRunner{
+			{Path: path.Join(buildPath, "metal", "bin", "ollama-runner")},
+			{Path: path.Join(buildPath, "cpu", "bin", "ollama-runner")},
 		}
 	case "linux":
-		runners = []string{
-			path.Join(buildPath, "cuda", "bin", "ollama-runner"),
-			path.Join(buildPath, "cpu", "bin", "ollama-runner"),
+		runners = []ModelRunner{
+			{Path: path.Join(buildPath, "cuda", "bin", "ollama-runner"), Accelerated: true},
+			{Path: path.Join(buildPath, "cpu", "bin", "ollama-runner")},
 		}
 	case "windows":
 		// TODO: select windows GPU runner here when available
-		runners = []string{
-			path.Join(buildPath, "cpu", "bin", "Release", "ollama-runner.exe"),
+		runners = []ModelRunner{
+			{Path: path.Join(buildPath, "cpu", "bin", "Release", "ollama-runner.exe")},
 		}
 	default:
 		log.Printf("unknown OS, running on CPU: %s", runtime.GOOS)
-		runners = []string{
-			path.Join(buildPath, "cpu", "bin", "ollama-runner"),
+		runners = []ModelRunner{
+			{Path: path.Join(buildPath, "cpu", "bin", "ollama-runner")},
 		}
 	}
 
 	runnerAvailable := false // if no runner files are found in the embed, this flag will cause a fast fail
 	for _, r := range runners {
 		// find all the files in the runner's bin directory
-		files, err := fs.Glob(llamaCppEmbed, path.Join(path.Dir(r), "*"))
+		files, err := fs.Glob(llamaCppEmbed, path.Join(path.Dir(r.Path), "*"))
 		if err != nil {
 			// this is expected, ollama may be compiled without all runners packed in
-			log.Printf("%s runner not found: %v", r, err)
+			log.Printf("%s runner not found: %v", r.Path, err)
 			continue
 		}
 
@@ -115,7 +117,10 @@ func chooseRunners(workDir, runnerType string) []ModelRunner {
 	localRunnersByPriority := []ModelRunner{}
 	for _, r := range runners {
 		// clean the ModelRunner paths so that they match the OS we are running on
-		localRunnersByPriority = append(localRunnersByPriority, ModelRunner{Path: filepath.Clean(path.Join(workDir, r))})
+		localRunnersByPriority = append(localRunnersByPriority, ModelRunner{
+			Path:        filepath.Clean(path.Join(workDir, r.Path)),
+			Accelerated: r.Accelerated,
+		})
 	}
 
 	return localRunnersByPriority
@@ -178,12 +183,12 @@ type llamaHyperparameters struct {
 }
 
 type Running struct {
-	Port     int
-	Cmd      *exec.Cmd
-	Cancel   context.CancelFunc
-	exitOnce sync.Once
-	exitCh   chan error // channel to receive the exit status of the subprocess
-	exitErr  error      // error returned by the subprocess
+	Port          int
+	Cmd           *exec.Cmd
+	Cancel        context.CancelFunc
+	exitOnce      sync.Once
+	exitCh        chan error // channel to receive the exit status of the subprocess
+	*StatusWriter            // captures error messages from the llama runner process
 }
 
 type llama struct {
@@ -193,7 +198,7 @@ type llama struct {
 
 var errNoGPU = errors.New("nvidia-smi command failed")
 
-// CheckVRAM returns the available VRAM in MiB on Linux machines with NVIDIA GPUs
+// CheckVRAM returns the free VRAM in bytes on Linux machines with NVIDIA GPUs
 func CheckVRAM() (int64, error) {
 	cmd := exec.Command("nvidia-smi", "--query-gpu=memory.free", "--format=csv,noheader,nounits")
 	var stdout bytes.Buffer
@@ -203,7 +208,7 @@ func CheckVRAM() (int64, error) {
 		return 0, errNoGPU
 	}
 
-	var free int64
+	var freeMiB int64
 	scanner := bufio.NewScanner(&stdout)
 	for scanner.Scan() {
 		line := scanner.Text()
@@ -212,10 +217,16 @@ func CheckVRAM() (int64, error) {
 			return 0, fmt.Errorf("failed to parse available VRAM: %v", err)
 		}
 
-		free += vram
+		freeMiB += vram
 	}
 
-	return free, nil
+	freeBytes := freeMiB * 1024 * 1024
+	if freeBytes < 2*format.GigaByte {
+		log.Printf("less than 2 GB VRAM available, falling back to CPU only")
+		freeMiB = 0
+	}
+
+	return freeBytes, nil
 }
 
 func NumGPU(numLayer, fileSizeBytes int64, opts api.Options) int {
@@ -223,7 +234,7 @@ func NumGPU(numLayer, fileSizeBytes int64, opts api.Options) int {
 		return opts.NumGPU
 	}
 	if runtime.GOOS == "linux" {
-		vramMib, err := CheckVRAM()
+		freeBytes, err := CheckVRAM()
 		if err != nil {
 			if err.Error() != "nvidia-smi command failed" {
 				log.Print(err.Error())
@@ -232,15 +243,13 @@ func NumGPU(numLayer, fileSizeBytes int64, opts api.Options) int {
 			return 0
 		}
 
-		freeVramBytes := int64(vramMib) * 1024 * 1024 // 1 MiB = 1024^2 bytes
-
 		// Calculate bytes per layer
 		// TODO: this is a rough heuristic, better would be to calculate this based on number of layers and context size
 		bytesPerLayer := fileSizeBytes / numLayer
 
-		// max number of layers we can fit in VRAM, subtract 5% to prevent consuming all available VRAM and running out of memory
-		layers := int(freeVramBytes/bytesPerLayer) * 95 / 100
-		log.Printf("%d MiB VRAM available, loading up to %d GPU layers", vramMib, layers)
+		// max number of layers we can fit in VRAM, subtract 8% to prevent consuming all available VRAM and running out of memory
+		layers := int(freeBytes/bytesPerLayer) * 92 / 100
+		log.Printf("%d MB VRAM available, loading up to %d GPU layers", freeBytes/(1024*1024), layers)
 
 		return layers
 	}
@@ -250,7 +259,8 @@ func NumGPU(numLayer, fileSizeBytes int64, opts api.Options) int {
 
 // StatusWriter is a writer that captures error messages from the llama runner process
 type StatusWriter struct {
-	ErrCh chan error
+	ErrCh      chan error
+	LastErrMsg string
 }
 
 func NewStatusWriter() *StatusWriter {
@@ -260,10 +270,18 @@ func NewStatusWriter() *StatusWriter {
 }
 
 func (w *StatusWriter) Write(b []byte) (int, error) {
+	var errMsg string
 	if _, after, ok := bytes.Cut(b, []byte("error:")); ok {
-		err := fmt.Errorf("llama runner: %s", after)
-		w.ErrCh <- err
+		errMsg = string(bytes.TrimSpace(after))
+	} else if _, after, ok := bytes.Cut(b, []byte("CUDA error")); ok {
+		errMsg = string(bytes.TrimSpace(after))
 	}
+
+	if errMsg != "" {
+		w.LastErrMsg = errMsg
+		w.ErrCh <- fmt.Errorf("llama runner: %s", errMsg)
+	}
+
 	return os.Stderr.Write(b)
 }
 
@@ -277,13 +295,14 @@ func newLlama(model string, adapters []string, runners []ModelRunner, numLayers
 		return nil, errors.New("ollama supports only one lora adapter, but multiple were provided")
 	}
 
+	numGPU := NumGPU(numLayers, fileInfo.Size(), opts)
 	params := []string{
 		"--model", model,
 		"--ctx-size", fmt.Sprintf("%d", opts.NumCtx),
 		"--rope-freq-base", fmt.Sprintf("%f", opts.RopeFrequencyBase),
 		"--rope-freq-scale", fmt.Sprintf("%f", opts.RopeFrequencyScale),
 		"--batch-size", fmt.Sprintf("%d", opts.NumBatch),
-		"--n-gpu-layers", fmt.Sprintf("%d", NumGPU(numLayers, fileInfo.Size(), opts)),
+		"--n-gpu-layers", fmt.Sprintf("%d", numGPU),
 		"--embedding",
 	}
 
@@ -317,6 +336,11 @@ func newLlama(model string, adapters []string, runners []ModelRunner, numLayers
 
 	// start the llama.cpp server with a retry in case the port is already in use
 	for _, runner := range runners {
+		if runner.Accelerated && numGPU == 0 {
+			log.Printf("skipping accelerated runner because num_gpu=0")
+			continue
+		}
+
 		if _, err := os.Stat(runner.Path); err != nil {
 			log.Printf("llama runner not found: %v", err)
 			continue
@@ -345,7 +369,13 @@ func newLlama(model string, adapters []string, runners []ModelRunner, numLayers
 		// monitor the llama runner process and signal when it exits
 		go func() {
 			err := llm.Cmd.Wait()
-			llm.exitErr = err
+			// default to printing the exit message of the command process, it will probably just say 'exit staus 1'
+			errMsg := err.Error()
+			// try to set a better error message if llama runner logs captured an error
+			if statusWriter.LastErrMsg != "" {
+				errMsg = statusWriter.LastErrMsg
+			}
+			log.Println(errMsg)
 			// llm.Cmd.Wait() can only be called once, use this exit channel to signal that the process has exited
 			llm.exitOnce.Do(func() {
 				close(llm.exitCh)
@@ -415,10 +445,9 @@ func (llm *llama) Close() {
 
 	// wait for the command to exit to prevent race conditions with the next run
 	<-llm.exitCh
-	err := llm.exitErr
 
-	if err != nil {
-		log.Printf("llama runner stopped with error: %v", err)
+	if llm.StatusWriter != nil && llm.StatusWriter.LastErrMsg != "" {
+		log.Printf("llama runner stopped with error: %v", llm.StatusWriter.LastErrMsg)
 	} else {
 		log.Print("llama runner stopped successfully")
 	}
@@ -428,71 +457,21 @@ func (llm *llama) SetOptions(opts api.Options) {
 	llm.Options = opts
 }
 
-type GenerationSettings struct {
-	FrequencyPenalty float64       `json:"frequency_penalty"`
-	IgnoreEOS        bool          `json:"ignore_eos"`
-	LogitBias        []interface{} `json:"logit_bias"`
-	Mirostat         int           `json:"mirostat"`
-	MirostatEta      float64       `json:"mirostat_eta"`
-	MirostatTau      float64       `json:"mirostat_tau"`
-	Model            string        `json:"model"`
-	NCtx             int           `json:"n_ctx"`
-	NKeep            int           `json:"n_keep"`
-	NPredict         int           `json:"n_predict"`
-	NProbs           int           `json:"n_probs"`
-	PenalizeNl       bool          `json:"penalize_nl"`
-	PresencePenalty  float64       `json:"presence_penalty"`
-	RepeatLastN      int           `json:"repeat_last_n"`
-	RepeatPenalty    float64       `json:"repeat_penalty"`
-	Seed             uint32        `json:"seed"`
-	Stop             []string      `json:"stop"`
-	Stream           bool          `json:"stream"`
-	Temp             float64       `json:"temp"`
-	TfsZ             float64       `json:"tfs_z"`
-	TopK             int           `json:"top_k"`
-	TopP             float64       `json:"top_p"`
-	TypicalP         float64       `json:"typical_p"`
-}
-
-type Timings struct {
-	PredictedN  int     `json:"predicted_n"`
-	PredictedMS float64 `json:"predicted_ms"`
-	PromptN     int     `json:"prompt_n"`
-	PromptMS    float64 `json:"prompt_ms"`
-}
-
-type Prediction struct {
+type prediction struct {
 	Content string `json:"content"`
 	Model   string `json:"model"`
 	Prompt  string `json:"prompt"`
 	Stop    bool   `json:"stop"`
 
-	Timings `json:"timings"`
+	Timings struct {
+		PredictedN  int     `json:"predicted_n"`
+		PredictedMS float64 `json:"predicted_ms"`
+		PromptN     int     `json:"prompt_n"`
+		PromptMS    float64 `json:"prompt_ms"`
+	}
 }
 
-type PredictRequest struct {
-	Prompt           string   `json:"prompt"`
-	Stream           bool     `json:"stream"`
-	NPredict         int      `json:"n_predict"`
-	NKeep            int      `json:"n_keep"`
-	Temperature      float32  `json:"temperature"`
-	TopK             int      `json:"top_k"`
-	TopP             float32  `json:"top_p"`
-	TfsZ             float32  `json:"tfs_z"`
-	TypicalP         float32  `json:"typical_p"`
-	RepeatLastN      int      `json:"repeat_last_n"`
-	RepeatPenalty    float32  `json:"repeat_penalty"`
-	PresencePenalty  float32  `json:"presence_penalty"`
-	FrequencyPenalty float32  `json:"frequency_penalty"`
-	Mirostat         int      `json:"mirostat"`
-	MirostatTau      float32  `json:"mirostat_tau"`
-	MirostatEta      float32  `json:"mirostat_eta"`
-	PenalizeNl       bool     `json:"penalize_nl"`
-	Seed             int      `json:"seed"`
-	Stop             []string `json:"stop,omitempty"`
-}
-
-const maxBufferSize = 512 * 1000 // 512KB
+const maxBufferSize = 512 * format.KiloByte
 
 func (llm *llama) Predict(ctx context.Context, prevContext []int, prompt string, fn func(api.GenerateResponse)) error {
 	prevConvo, err := llm.Decode(ctx, prevContext)
@@ -500,39 +479,46 @@ func (llm *llama) Predict(ctx context.Context, prevContext []int, prompt string,
 		return err
 	}
 
+	// Remove leading spaces from prevConvo if present
+	prevConvo = strings.TrimPrefix(prevConvo, " ")
+
 	var nextContext strings.Builder
 	nextContext.WriteString(prevConvo)
 	nextContext.WriteString(prompt)
 
+	request := map[string]any{
+		"prompt":            nextContext.String(),
+		"stream":            true,
+		"n_predict":         llm.NumPredict,
+		"n_keep":            llm.NumKeep,
+		"temperature":       llm.Temperature,
+		"top_k":             llm.TopK,
+		"top_p":             llm.TopP,
+		"tfs_z":             llm.TFSZ,
+		"typical_p":         llm.TypicalP,
+		"repeat_last_n":     llm.RepeatLastN,
+		"repeat_penalty":    llm.RepeatPenalty,
+		"presence_penalty":  llm.PresencePenalty,
+		"frequency_penalty": llm.FrequencyPenalty,
+		"mirostat":          llm.Mirostat,
+		"mirostat_tau":      llm.MirostatTau,
+		"mirostat_eta":      llm.MirostatEta,
+		"penalize_nl":       llm.PenalizeNewline,
+		"seed":              llm.Seed,
+		"stop":              llm.Stop,
+	}
+
+	// Handling JSON marshaling with special characters unescaped.
+	buffer := &bytes.Buffer{}
+	enc := json.NewEncoder(buffer)
+	enc.SetEscapeHTML(false)
+
+	if err := enc.Encode(request); err != nil {
+		return fmt.Errorf("failed to marshal data: %v", err)
+	}
+
 	endpoint := fmt.Sprintf("http://127.0.0.1:%d/completion", llm.Port)
-	predReq := PredictRequest{
-		Prompt:           nextContext.String(),
-		Stream:           true,
-		NPredict:         llm.NumPredict,
-		NKeep:            llm.NumKeep,
-		Temperature:      llm.Temperature,
-		TopK:             llm.TopK,
-		TopP:             llm.TopP,
-		TfsZ:             llm.TFSZ,
-		TypicalP:         llm.TypicalP,
-		RepeatLastN:      llm.RepeatLastN,
-		RepeatPenalty:    llm.RepeatPenalty,
-		PresencePenalty:  llm.PresencePenalty,
-		FrequencyPenalty: llm.FrequencyPenalty,
-		Mirostat:         llm.Mirostat,
-		MirostatTau:      llm.MirostatTau,
-		MirostatEta:      llm.MirostatEta,
-		PenalizeNl:       llm.PenalizeNewline,
-		Seed:             llm.Seed,
-		Stop:             llm.Stop,
-	}
-
-	data, err := json.Marshal(predReq)
-	if err != nil {
-		return fmt.Errorf("error marshaling data: %v", err)
-	}
-
-	req, err := http.NewRequestWithContext(ctx, http.MethodPost, endpoint, bytes.NewBuffer(data))
+	req, err := http.NewRequestWithContext(ctx, http.MethodPost, endpoint, buffer)
 	if err != nil {
 		return fmt.Errorf("error creating POST request: %v", err)
 	}
@@ -563,16 +549,14 @@ func (llm *llama) Predict(ctx context.Context, prevContext []int, prompt string,
 			// This handles the request cancellation
 			return ctx.Err()
 		default:
-			line := scanner.Text()
-			if line == "" {
+			line := scanner.Bytes()
+			if len(line) == 0 {
 				continue
 			}
 
-			// Read data from the server-side event stream
-			if strings.HasPrefix(line, "data: ") {
-				evt := line[6:]
-				var p Prediction
-				if err := json.Unmarshal([]byte(evt), &p); err != nil {
+			if evt, ok := bytes.CutPrefix(line, []byte("data: ")); ok {
+				var p prediction
+				if err := json.Unmarshal(evt, &p); err != nil {
 					return fmt.Errorf("error unmarshaling llm prediction response: %v", err)
 				}
 
@@ -590,10 +574,10 @@ func (llm *llama) Predict(ctx context.Context, prevContext []int, prompt string,
 					fn(api.GenerateResponse{
 						Done:               true,
 						Context:            embd,
-						PromptEvalCount:    p.PromptN,
-						PromptEvalDuration: parseDurationMs(p.PromptMS),
-						EvalCount:          p.PredictedN,
-						EvalDuration:       parseDurationMs(p.PredictedMS),
+						PromptEvalCount:    p.Timings.PromptN,
+						PromptEvalDuration: parseDurationMs(p.Timings.PromptMS),
+						EvalCount:          p.Timings.PredictedN,
+						EvalDuration:       parseDurationMs(p.Timings.PredictedMS),
 					})
 
 					return nil
@@ -603,6 +587,14 @@ func (llm *llama) Predict(ctx context.Context, prevContext []int, prompt string,
 	}
 
 	if err := scanner.Err(); err != nil {
+		if strings.Contains(err.Error(), "unexpected EOF") {
+			// this means the llama runner subprocess crashed
+			llm.Close()
+			if llm.StatusWriter != nil && llm.StatusWriter.LastErrMsg != "" {
+				return fmt.Errorf("llama runner exited: %v", llm.StatusWriter.LastErrMsg)
+			}
+			return fmt.Errorf("llama runner exited, you may not have enough available memory to run this model")
+		}
 		return fmt.Errorf("error reading llm response: %v", err)
 	}
 
@@ -699,9 +691,6 @@ func (llm *llama) Decode(ctx context.Context, tokens []int) (string, error) {
 		return "", fmt.Errorf("unmarshal encode response: %w", err)
 	}
 
-	// decoded content contains a leading whitespace
-	decoded.Content, _ = strings.CutPrefix(decoded.Content, "")
-
 	return decoded.Content, nil
 }
 

llm/llm.go

@@ -10,6 +10,7 @@ import (
 	"github.com/pbnjay/memory"
 
 	"github.com/jmorganca/ollama/api"
+	"github.com/jmorganca/ollama/format"
 )
 
 type LLM interface {
@@ -55,39 +56,30 @@ func New(workDir, model string, adapters []string, opts api.Options) (LLM, error
 				opts.NumGPU = 0
 			}
 		}
-	}
 
-	totalResidentMemory := memory.TotalMemory()
-	switch ggml.ModelType() {
-	case "3B", "7B":
-		if ggml.FileType() == "F16" && totalResidentMemory < 16*1000*1000 {
-			return nil, fmt.Errorf("F16 model requires at least 16 GB of memory")
-		} else if totalResidentMemory < 8*1000*1000 {
-			return nil, fmt.Errorf("model requires at least 8 GB of memory")
+		var requiredMemory int64
+		var f16Multiplier int64 = 2
+
+		switch ggml.ModelType() {
+		case "3B", "7B":
+			requiredMemory = 8 * format.GigaByte
+		case "13B":
+			requiredMemory = 16 * format.GigaByte
+		case "30B", "34B", "40B":
+			requiredMemory = 32 * format.GigaByte
+		case "65B", "70B":
+			requiredMemory = 64 * format.GigaByte
+		case "180B":
+			requiredMemory = 128 * format.GigaByte
+			f16Multiplier = 4
 		}
-	case "13B":
-		if ggml.FileType() == "F16" && totalResidentMemory < 32*1000*1000 {
-			return nil, fmt.Errorf("F16 model requires at least 32 GB of memory")
-		} else if totalResidentMemory < 16*1000*1000 {
-			return nil, fmt.Errorf("model requires at least 16 GB of memory")
-		}
-	case "30B", "34B", "40B":
-		if ggml.FileType() == "F16" && totalResidentMemory < 64*1000*1000 {
-			return nil, fmt.Errorf("F16 model requires at least 64 GB of memory")
-		} else if totalResidentMemory < 32*1000*1000 {
-			return nil, fmt.Errorf("model requires at least 32 GB of memory")
-		}
-	case "65B", "70B":
-		if ggml.FileType() == "F16" && totalResidentMemory < 128*1000*1000 {
-			return nil, fmt.Errorf("F16 model requires at least 128 GB of memory")
-		} else if totalResidentMemory < 64*1000*1000 {
-			return nil, fmt.Errorf("model requires at least 64 GB of memory")
-		}
-	case "180B":
-		if ggml.FileType() == "F16" && totalResidentMemory < 512*1000*1000 {
-			return nil, fmt.Errorf("F16 model requires at least 512GB of memory")
-		} else if totalResidentMemory < 128*1000*1000 {
-			return nil, fmt.Errorf("model requires at least 128GB of memory")
+
+		systemMemory := int64(memory.TotalMemory())
+
+		if ggml.FileType() == "F16" && requiredMemory*f16Multiplier > systemMemory {
+			return nil, fmt.Errorf("F16 model requires at least %s of total memory", format.HumanBytes(requiredMemory))
+		} else if requiredMemory > systemMemory {
+			return nil, fmt.Errorf("model requires at least %s of total memory", format.HumanBytes(requiredMemory))
 		}
 	}
 

parser/parser.go

@@ -40,7 +40,7 @@ func Parse(reader io.Reader) ([]Command, error) {
 			command.Args = string(fields[1])
 			// copy command for validation
 			modelCommand = command
-		case "LICENSE", "TEMPLATE", "SYSTEM", "PROMPT", "EMBED", "ADAPTER":
+		case "LICENSE", "TEMPLATE", "SYSTEM", "PROMPT", "ADAPTER":
 			command.Name = string(bytes.ToLower(fields[0]))
 			command.Args = string(fields[1])
 		case "PARAMETER":
@@ -51,6 +51,8 @@ func Parse(reader io.Reader) ([]Command, error) {
 
 			command.Name = string(fields[0])
 			command.Args = string(fields[1])
+		case "EMBED":
+			return nil, fmt.Errorf("deprecated command: EMBED is no longer supported, use the /embed API endpoint instead")
 		default:
 			if !bytes.HasPrefix(fields[0], []byte("#")) {
 				// log a warning for unknown commands

runner/.gitignore

@@ -0,0 +1,2 @@
+model.bin
+runner

runner/darwin.go

@@ -0,0 +1,39 @@
+package main
+
+import (
+	"embed"
+	"io"
+	"os"
+	"path/filepath"
+)
+
+//go:embed ggml-metal.metal
+var fs embed.FS
+
+func init() {
+	exec, err := os.Executable()
+	if err != nil {
+		return
+	}
+
+	exec, err = filepath.EvalSymlinks(exec)
+	if err != nil {
+		return
+	}
+
+	dst, err := os.Create(filepath.Join(filepath.Dir(exec), "ggml-metal.metal"))
+	if err != nil {
+		return
+	}
+	defer dst.Close()
+
+	src, err := fs.Open("ggml-metal.metal")
+	if err != nil {
+		return
+	}
+	defer src.Close()
+
+	if _, err := io.Copy(dst, src); err != nil {
+		return
+	}
+}

runner/ggml-alloc.c

@@ -0,0 +1,620 @@
+/**
+ * llama.cpp - git 465219b9143ac01db0990bbcb0a081ef72ec2008
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023 Georgi Gerganov
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "ggml-alloc.h"
+#include "ggml-backend.h"
+#include "ggml.h"
+#include <assert.h>
+#include <stdarg.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+
+#define UNUSED(x) (void)(x)
+#define MAX(a, b) ((a) > (b) ? (a) : (b))
+#define GGML_MAX_CONCUR (2*GGML_MAX_NODES)
+
+//#define GGML_ALLOCATOR_DEBUG
+
+//#define AT_PRINTF printf
+#define AT_PRINTF(...) ((void)0)
+
+struct hash_node {
+    struct ggml_tensor * t;
+    int n_children;
+    int n_views;
+};
+
+static size_t hash(void * p) {
+    return (size_t)p % GGML_GRAPH_HASHTABLE_SIZE;
+}
+
+static struct hash_node * hash_get(struct hash_node hash_table[], struct ggml_tensor * t) {
+    size_t h = hash(t);
+
+    // linear probing
+    size_t i = h;
+    while (hash_table[i].t != NULL) {
+        if (hash_table[i].t == t) {
+            return &hash_table[i];
+        }
+        i = (i + 1) % GGML_GRAPH_HASHTABLE_SIZE;
+        if (i == h) {
+            // hash table is full
+            GGML_ASSERT(false);
+        }
+    }
+
+    hash_table[i].t = t;
+    return &hash_table[i];
+}
+
+// TODO: GGML_PAD ?
+static size_t aligned_offset(const void * buffer, size_t offset, size_t alignment) {
+    assert(alignment && !(alignment & (alignment - 1))); // power of 2
+    size_t align = (alignment - (((uintptr_t)buffer + offset) % alignment)) % alignment;
+    return offset + align;
+}
+
+struct free_block {
+    void * addr;
+    size_t size;
+};
+
+#define MAX_FREE_BLOCKS 256
+
+struct ggml_allocr {
+    struct ggml_backend_buffer * buffer;
+    bool buffer_owned;
+    void * data;
+    size_t alignment;
+    int n_free_blocks;
+    struct free_block free_blocks[MAX_FREE_BLOCKS];
+    struct hash_node hash_table[GGML_GRAPH_HASHTABLE_SIZE];
+    size_t max_size;
+    bool measure;
+    int parse_seq[GGML_MAX_CONCUR];
+    int parse_seq_len;
+
+#ifdef GGML_ALLOCATOR_DEBUG
+    struct ggml_tensor * allocated_tensors[1024];
+#endif
+};
+
+#ifdef GGML_ALLOCATOR_DEBUG
+static void add_allocated_tensor(struct ggml_allocr * alloc, struct ggml_tensor * tensor) {
+    for (int i = 0; i < 1024; i++) {
+        if (alloc->allocated_tensors[i] == NULL) {
+            alloc->allocated_tensors[i] = tensor;
+            return;
+        }
+    }
+    GGML_ASSERT(!"out of allocated_tensors");
+}
+static void remove_allocated_tensor(struct ggml_allocr * alloc, struct ggml_tensor * tensor) {
+    for (int i = 0; i < 1024; i++) {
+        if (alloc->allocated_tensors[i] == tensor ||
+            (alloc->allocated_tensors[i] != NULL && alloc->allocated_tensors[i]->data == tensor->data)) {
+            alloc->allocated_tensors[i] = NULL;
+            return;
+        }
+    }
+    printf("tried to free tensor %s not found\n", tensor->name);
+    GGML_ASSERT(!"tensor not found");
+}
+#endif
+
+// check if a tensor is allocated by this buffer
+static bool ggml_allocr_is_own(struct ggml_allocr * alloc, const struct ggml_tensor * tensor) {
+    return tensor->buffer == alloc->buffer;
+}
+
+static bool ggml_is_view(struct ggml_tensor * t) {
+    return t->view_src != NULL;
+}
+
+void ggml_allocr_alloc(struct ggml_allocr * alloc, struct ggml_tensor * tensor) {
+    GGML_ASSERT(!ggml_is_view(tensor)); // views generally get data pointer from one of their sources
+    GGML_ASSERT(tensor->data == NULL); // avoid allocating tensor which already has memory allocated
+
+    size_t size = ggml_backend_buffer_get_alloc_size(alloc->buffer, tensor);
+    size = aligned_offset(NULL, size, alloc->alignment);
+
+    AT_PRINTF("%s: allocating %s (%zu bytes) - ", __func__, tensor->name, size);
+
+    size_t max_avail = 0;
+
+    // find the best fitting free block besides the last block
+    int best_fit_block = -1;
+    size_t best_fit_size = SIZE_MAX;
+    for (int i = 0; i < alloc->n_free_blocks - 1; i++) {
+        struct free_block * block = &alloc->free_blocks[i];
+        max_avail = MAX(max_avail, block->size);
+        if (block->size >= size && block->size <= best_fit_size) {
+            best_fit_block = i;
+            best_fit_size = block->size;
+        }
+    }
+
+    AT_PRINTF("block %d\n", best_fit_block);
+
+    if (best_fit_block == -1) {
+        // the last block is our last resort
+        struct free_block * block = &alloc->free_blocks[alloc->n_free_blocks - 1];
+        max_avail = MAX(max_avail, block->size);
+        if (block->size >= size) {
+            best_fit_block = alloc->n_free_blocks - 1;
+        } else {
+            fprintf(stderr, "%s: not enough space in the buffer (needed %zu, largest block available %zu)\n",
+                    __func__, size, max_avail);
+            GGML_ASSERT(!"not enough space in the buffer");
+            return;
+        }
+    }
+    struct free_block * block = &alloc->free_blocks[best_fit_block];
+    void * addr = block->addr;
+    block->addr = (char*)block->addr + size;
+    block->size -= size;
+    if (block->size == 0) {
+        // remove block if empty
+        alloc->n_free_blocks--;
+        for (int j = best_fit_block; j < alloc->n_free_blocks; j++) {
+            alloc->free_blocks[j] = alloc->free_blocks[j+1];
+        }
+    }
+
+    tensor->data = addr;
+    AT_PRINTF("%s: allocated data at %p\n", __func__, tensor->data);
+    tensor->buffer = alloc->buffer;
+    ggml_backend_buffer_init_tensor(alloc->buffer, tensor);
+
+#ifdef GGML_ALLOCATOR_DEBUG
+    add_allocated_tensor(alloc, tensor);
+    size_t cur_max = (char*)addr - (char*)alloc->data + size;
+    if (cur_max > alloc->max_size) {
+        printf("max_size = %.2f MB: tensors: ", cur_max / 1024.0 / 1024.0);
+        for (int i = 0; i < 1024; i++) {
+            if (alloc->allocated_tensors[i]) {
+                printf("%s (%.2f MB) ", alloc->allocated_tensors[i]->name, ggml_nbytes(alloc->allocated_tensors[i]) / 1024.0 / 1024.0);
+            }
+        }
+        printf("\n");
+    }
+#endif
+
+    alloc->max_size = MAX(alloc->max_size, (char*)addr - (char*)alloc->data + size);
+}
+
+// this is a very naive implementation, but for our case the number of free blocks should be very small
+static void ggml_allocr_free_tensor(struct ggml_allocr * alloc, struct ggml_tensor * tensor) {
+    if (ggml_allocr_is_own(alloc, tensor) == false) {
+        // the tensor was not allocated in this buffer
+        // this can happen because the graph allocator will try to free weights and other tensors from different buffers
+        // the easiest way to deal with this is just to ignore it
+        AT_PRINTF("ignoring %s (their buffer: %p, our buffer: %p)\n", tensor->name, (void *)tensor->buffer, (void *)alloc->buffer);
+        return;
+    }
+
+    void * ptr = tensor->data;
+
+    size_t size = ggml_backend_buffer_get_alloc_size(alloc->buffer, tensor);
+    size = aligned_offset(NULL, size, alloc->alignment);
+    AT_PRINTF("%s: freeing %s at %p (%zu bytes) - n_free_blocks = %d\n", __func__, tensor->name, ptr, size, alloc->n_free_blocks);
+
+    ggml_backend_buffer_free_tensor(alloc->buffer, tensor);
+
+#ifdef GGML_ALLOCATOR_DEBUG
+    remove_allocated_tensor(alloc, tensor);
+#endif
+
+    // see if we can merge with an existing block
+    for (int i = 0; i < alloc->n_free_blocks; i++) {
+        struct free_block * block = &alloc->free_blocks[i];
+        // check if ptr is at the end of the block
+        if ((char*)block->addr + block->size == ptr) {
+            block->size += size;
+            // check if we can merge with the next block
+            if (i < alloc->n_free_blocks - 1 && (char*)block->addr + block->size == alloc->free_blocks[i+1].addr) {
+                block->size += alloc->free_blocks[i+1].size;
+                alloc->n_free_blocks--;
+                for (int j = i+1; j < alloc->n_free_blocks; j++) {
+                    alloc->free_blocks[j] = alloc->free_blocks[j+1];
+                }
+            }
+            return;
+        }
+        // check if ptr is at the beginning of the block
+        if ((char*)ptr + size == block->addr) {
+            block->addr = ptr;
+            block->size += size;
+            // check if we can merge with the previous block
+            if (i > 0 && (char*)alloc->free_blocks[i-1].addr + alloc->free_blocks[i-1].size == block->addr) {
+                alloc->free_blocks[i-1].size += block->size;
+                alloc->n_free_blocks--;
+                for (int j = i; j < alloc->n_free_blocks; j++) {
+                    alloc->free_blocks[j] = alloc->free_blocks[j+1];
+                }
+            }
+            return;
+        }
+    }
+    // otherwise, add a new block
+    GGML_ASSERT(alloc->n_free_blocks < MAX_FREE_BLOCKS && "out of free blocks");
+    // insert the new block in the correct position to keep the array sorted by address (to make merging blocks faster)
+    int insert_pos = 0;
+    while (insert_pos < alloc->n_free_blocks && alloc->free_blocks[insert_pos].addr < ptr) {
+        insert_pos++;
+    }
+    // shift all blocks from insert_pos onward to make room for the new block
+    for (int i = alloc->n_free_blocks; i > insert_pos; i--) {
+        alloc->free_blocks[i] = alloc->free_blocks[i-1];
+    }
+    // insert the new block
+    alloc->free_blocks[insert_pos].addr = ptr;
+    alloc->free_blocks[insert_pos].size = size;
+    alloc->n_free_blocks++;
+}
+
+void ggml_allocr_set_parse_seq(struct ggml_allocr * alloc, const int * list, int n) {
+    for (int i = 0; i < n; i++) {
+        alloc->parse_seq[i] = list[i];
+    }
+    alloc->parse_seq_len = n;
+}
+
+void ggml_allocr_reset(struct ggml_allocr * alloc) {
+    alloc->n_free_blocks = 1;
+    size_t align_offset = aligned_offset(alloc->data, 0, alloc->alignment);
+    alloc->free_blocks[0].addr = (char *)alloc->data + align_offset;
+    alloc->free_blocks[0].size = ggml_backend_buffer_get_size(alloc->buffer) - align_offset;
+}
+
+struct ggml_allocr * ggml_allocr_new(void * data, size_t size, size_t alignment) {
+    struct ggml_backend_buffer * buffer = ggml_backend_cpu_buffer_from_ptr(NULL, data, size);
+
+    struct ggml_allocr * alloc = (struct ggml_allocr *)malloc(sizeof(struct ggml_allocr));
+
+    *alloc = (struct ggml_allocr){
+        /*.buffer        = */ buffer,
+        /*.buffer_owned  = */ true,
+        /*.base          = */ ggml_backend_buffer_get_base(buffer),
+        /*.alignment     = */ alignment,
+        /*.n_free_blocks = */ 0,
+        /*.free_blocks   = */ {{0}},
+        /*.hash_table    = */ {{0}},
+        /*.max_size      = */ 0,
+        /*.measure       = */ false,
+        /*.parse_seq     = */ {0},
+        /*.parse_seq_len = */ 0,
+#ifdef GGML_ALLOCATOR_DEBUG
+        /*.allocated_tensors = */ {0},
+#endif
+    };
+
+    ggml_allocr_reset(alloc);
+
+    return alloc;
+}
+
+struct ggml_allocr * ggml_allocr_new_measure(size_t alignment) {
+    struct ggml_allocr * alloc = ggml_allocr_new((void *)0x1000, (size_t)-0x1001, alignment);
+    alloc->measure = true;
+
+    return alloc;
+}
+
+struct ggml_allocr * ggml_allocr_new_from_buffer(struct ggml_backend_buffer * buffer) {
+    struct ggml_allocr * alloc = (struct ggml_allocr *)malloc(sizeof(struct ggml_allocr));
+
+    *alloc = (struct ggml_allocr){
+        /*.buffer        = */ buffer,
+        /*.buffer_owned  = */ false,
+        /*.base          = */ ggml_backend_buffer_get_base(buffer),
+        /*.alignment     = */ ggml_backend_buffer_get_alignment(buffer),
+        /*.n_free_blocks = */ 0,
+        /*.free_blocks   = */ {{0}},
+        /*.hash_table    = */ {{0}},
+        /*.max_size      = */ 0,
+        /*.measure       = */ false,
+        /*.parse_seq     = */ {0},
+        /*.parse_seq_len = */ 0,
+#ifdef GGML_ALLOCATOR_DEBUG
+        /*.allocated_tensors = */ {0},
+#endif
+    };
+
+    ggml_allocr_reset(alloc);
+
+    return alloc;
+}
+
+void ggml_allocr_free(struct ggml_allocr * alloc) {
+    if (alloc->buffer_owned) {
+        ggml_backend_buffer_free(alloc->buffer);
+    }
+    free(alloc);
+}
+
+bool ggml_allocr_is_measure(struct ggml_allocr * alloc) {
+    return alloc->measure;
+}
+
+//////////// compute graph allocator
+
+static bool ggml_are_same_layout(const struct ggml_tensor * a, const struct ggml_tensor * b) {
+    if (a->type != b->type) {
+        return false;
+    }
+    for (int i = 0; i < GGML_MAX_DIMS; i++) {
+        if (a->ne[i] != b->ne[i]) {
+            return false;
+        }
+        if (a->nb[i] != b->nb[i]) {
+            return false;
+        }
+    }
+    return true;
+}
+
+static bool ggml_op_can_inplace(enum ggml_op op) {
+    switch (op) {
+        case GGML_OP_SCALE:
+        case GGML_OP_DIAG_MASK_ZERO:
+        case GGML_OP_DIAG_MASK_INF:
+        case GGML_OP_ADD:
+        case GGML_OP_ADD1:
+        case GGML_OP_SUB:
+        case GGML_OP_MUL:
+        case GGML_OP_DIV:
+        case GGML_OP_SQR:
+        case GGML_OP_SQRT:
+        case GGML_OP_LOG:
+        case GGML_OP_UNARY:
+        case GGML_OP_ROPE:
+        case GGML_OP_RMS_NORM:
+        case GGML_OP_SOFT_MAX:
+            return true;
+
+        default:
+            return false;
+    }
+}
+
+static void init_view(struct ggml_allocr * alloc, struct ggml_tensor * view) {
+    assert(view->view_src != NULL && view->view_src->data != NULL);
+    view->backend = view->view_src->backend;
+    view->buffer  = view->view_src->buffer;
+    view->data    = (char *)view->view_src->data + view->view_offs;
+
+    // FIXME: the view should be initialized by the owning buffer, but currently this breaks the CUDA backend
+    // due to the ggml_tensor_extra_gpu ring buffer overwriting the KV cache extras
+    assert(ggml_allocr_is_measure(alloc) || !view->buffer || view->buffer->backend == alloc->buffer->backend);
+    ggml_backend_buffer_init_tensor(alloc->buffer, view);
+}
+
+static void allocate_node(struct ggml_allocr * alloc, struct ggml_tensor * node) {
+    struct hash_node * ht = alloc->hash_table;
+    if (node->data == NULL) {
+        if (ggml_is_view(node)) {
+            init_view(alloc, node);
+        } else {
+            // see if we can reuse a parent's buffer (inplace)
+            if (ggml_op_can_inplace(node->op)) {
+                for (int i = 0; i < GGML_MAX_SRC; i++) {
+                    struct ggml_tensor * parent = node->src[i];
+                    if (parent == NULL) {
+                        break;
+                    }
+
+                    // if the node's data is external, then we cannot re-use it
+                    if (ggml_allocr_is_own(alloc, parent) == false) {
+                        AT_PRINTF("not reusing parent %s for %s as %p is external\n", parent->name, node->name, parent->data);
+                        continue;
+                    }
+
+                    struct hash_node * p_hn = hash_get(ht, parent);
+                    if (parent->data != NULL && p_hn->n_children == 1 && p_hn->n_views == 0 && ggml_are_same_layout(node, parent)) {
+                        if (ggml_is_view(parent)) {
+                            struct ggml_tensor * view_src = parent->view_src;
+                            struct hash_node * view_src_hn = hash_get(ht, view_src);
+                            if (view_src_hn->n_views == 1 && view_src_hn->n_children == 0 && view_src->data == parent->data) {
+                                // TODO: the offset of the view parent must be kept to ensure that the op doesn't overwrite
+                                // the parent's data that it will need later (same layout requirement). the problem is that then
+                                // we cannot free the tensor because the original address of the allocation is lost.
+                                // adding a view_src pointer to the tensor would solve this and simplify the code dealing with views
+                                // for now, we only reuse the parent's data if the offset is zero (view_src->data == parent->data)
+                                AT_PRINTF("reusing view parent %s (%s) for %s\n", parent->name, view_src->name, node->name);
+                                node->view_src = view_src;
+                                view_src_hn->n_views += 1;
+                                init_view(alloc, node);
+                                return;
+                            }
+                        }
+                        else {
+                            AT_PRINTF("reusing parent %s for %s\n", parent->name, node->name);
+                            node->view_src = parent;
+                            p_hn->n_views += 1;
+                            init_view(alloc, node);
+                            return;
+                        }
+                    }
+                }
+            }
+            ggml_allocr_alloc(alloc, node);
+        }
+    }
+}
+
+size_t ggml_allocr_alloc_graph_n(
+    struct ggml_allocr * alloc,
+    struct ggml_cgraph ** graphs, int n_graphs,
+    struct ggml_tensor *** inputs, struct ggml_tensor *** outputs) {
+
+    // reset hash table
+    struct hash_node * ht = alloc->hash_table;
+    memset(ht, 0, sizeof(struct hash_node) * GGML_GRAPH_HASHTABLE_SIZE);
+
+    // count number of children and views
+    for (int g = 0; g < n_graphs; g++) {
+        struct ggml_cgraph * gf = graphs[g];
+        for (int i = 0; i < gf->n_nodes; i++) {
+            struct ggml_tensor * node = gf->nodes[i];
+
+            if (ggml_is_view(node)) {
+                struct ggml_tensor * view_src = node->view_src;
+                hash_get(ht, view_src)->n_views += 1;
+                if (node->buffer == NULL && node->data != NULL) {
+                    // view of a pre-allocated tensor, didn't call init_view() yet
+                    init_view(alloc, node);
+                }
+            }
+
+            for (int j = 0; j < GGML_MAX_SRC; j++) {
+                struct ggml_tensor * parent = node->src[j];
+                if (parent == NULL) {
+                    break;
+                }
+                hash_get(ht, parent)->n_children += 1;
+                if (ggml_is_view(parent) && parent->buffer == NULL && parent->data != NULL) {
+                    init_view(alloc, parent);
+                }
+            }
+        }
+    }
+
+    // allocate tensors
+    for (int g = 0; g < n_graphs; g++) {
+        struct ggml_cgraph * gf = graphs[g];
+        AT_PRINTF("####### graph %d/%d\n", g, n_graphs);
+        // graph inputs are allocated first to ensure that they are not overwritten by each other
+        if (inputs != NULL && inputs[g] != NULL) {
+            for (int i = 0; inputs[g][i] != NULL; i++) {
+                struct ggml_tensor * input = inputs[g][i];
+                AT_PRINTF("input: %s\n", input->name);
+                allocate_node(alloc, input);
+            }
+        }
+        // if we have parse_seq then we allocate nodes following the list, and we only free nodes at barriers
+        int last_barrier_pos = 0;
+        int n_nodes = alloc->parse_seq_len ? alloc->parse_seq_len : gf->n_nodes;
+
+        for (int ind = 0; ind < n_nodes; ind++) {
+            // allocate a node if there is no parse_seq or this is not a barrier
+            if ((alloc->parse_seq_len==0) || alloc->parse_seq[ind] != -1) {
+                int i = alloc->parse_seq_len ? alloc->parse_seq[ind] : ind;
+                struct ggml_tensor * node = gf->nodes[i];
+
+                // allocate parents (leafs)
+                for (int j = 0; j < GGML_MAX_SRC; j++) {
+                    struct ggml_tensor * parent = node->src[j];
+                    if (parent == NULL) {
+                        break;
+                    }
+                    allocate_node(alloc, parent);
+                }
+
+                // allocate node
+                allocate_node(alloc, node);
+
+                AT_PRINTF("exec: %s (%s) <= ", ggml_op_name(node->op), node->name);
+                for (int j = 0; j < GGML_MAX_SRC; j++) {
+                    struct ggml_tensor * parent = node->src[j];
+                    if (parent == NULL) {
+                        break;
+                    }
+                    AT_PRINTF("%s", parent->name);
+                    if (j < GGML_MAX_SRC - 1 && node->src[j + 1] != NULL) {
+                        AT_PRINTF(", ");
+                    }
+                }
+                AT_PRINTF("\n");
+            }
+
+            // update parents
+            // update immediately if there is no parse_seq
+            // update only at barriers if there is parse_seq
+            if ((alloc->parse_seq_len == 0) || alloc->parse_seq[ind] == -1) {
+                int update_start = alloc->parse_seq_len ? last_barrier_pos : ind;
+                int update_end   = alloc->parse_seq_len ? ind              : ind + 1;
+                for (int i = update_start; i < update_end; i++) {
+                    int node_i = alloc->parse_seq_len ? alloc->parse_seq[i] : i;
+                    struct ggml_tensor * node = gf->nodes[node_i];
+
+                    for (int j = 0; j < GGML_MAX_SRC; j++) {
+                        struct ggml_tensor * parent = node->src[j];
+                        if (parent == NULL) {
+                            break;
+                        }
+                        struct hash_node * p_hn = hash_get(ht, parent);
+                        p_hn->n_children -= 1;
+
+                        //AT_PRINTF("parent %s: %d children, %d views\n", parent->name, parent->n_children, parent->n_views);
+
+                        if (p_hn->n_children == 0 && p_hn->n_views == 0) {
+                            if (ggml_is_view(parent)) {
+                                struct ggml_tensor * view_src = parent->view_src;
+                                struct hash_node * view_src_hn = hash_get(ht, view_src);
+                                view_src_hn->n_views -= 1;
+                                AT_PRINTF("view_src %s: %d children, %d views\n", view_src->name, view_src_hn->n_children, view_src_hn->n_views);
+                                if (view_src_hn->n_views == 0 && view_src_hn->n_children == 0 && view_src->data != node->data) {
+                                    ggml_allocr_free_tensor(alloc, view_src);
+                                }
+                            }
+                            else {
+                                if (parent->data != node->data) {
+                                    ggml_allocr_free_tensor(alloc, parent);
+                                }
+                            }
+                        }
+                    }
+                }
+                AT_PRINTF("\n");
+                if (alloc->parse_seq_len) {
+                    last_barrier_pos = ind + 1;
+                }
+            }
+        }
+        // free graph outputs here that wouldn't be freed otherwise because they have no children
+        if (outputs != NULL && outputs[g] != NULL) {
+            for (int i = 0; outputs[g][i] != NULL; i++) {
+                struct ggml_tensor * output = outputs[g][i];
+                AT_PRINTF("output: %s\n", output->name);
+                ggml_allocr_free_tensor(alloc, output);
+            }
+        }
+    }
+
+    return alloc->max_size;
+}
+
+size_t ggml_allocr_alloc_graph(struct ggml_allocr * alloc, struct ggml_cgraph * graph) {
+    return ggml_allocr_alloc_graph_n(alloc, &graph, 1, NULL, NULL);
+}
+
+size_t ggml_allocr_max_size(struct ggml_allocr * alloc) {
+    return alloc->max_size;
+}

runner/ggml-alloc.h

@@ -0,0 +1,59 @@
+/**
+ * llama.cpp - git 465219b9143ac01db0990bbcb0a081ef72ec2008
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023 Georgi Gerganov
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#pragma once
+
+#include "ggml.h"
+
+#ifdef  __cplusplus
+extern "C" {
+#endif
+
+struct ggml_backend_buffer;
+
+GGML_API struct ggml_allocr * ggml_allocr_new(void * data, size_t size, size_t alignment);
+GGML_API struct ggml_allocr * ggml_allocr_new_measure(size_t alignment);
+GGML_API struct ggml_allocr * ggml_allocr_new_from_buffer(struct ggml_backend_buffer * buffer);
+
+// tell the allocator to parse nodes following the order described in the list
+// you should call this if your graph are optimized to execute out-of-order
+GGML_API void   ggml_allocr_set_parse_seq(struct ggml_allocr * alloc, const int * list, int n);
+
+GGML_API void   ggml_allocr_free       (struct ggml_allocr * alloc);
+GGML_API bool   ggml_allocr_is_measure (struct ggml_allocr * alloc);
+GGML_API void   ggml_allocr_reset      (struct ggml_allocr * alloc);
+GGML_API void   ggml_allocr_alloc      (struct ggml_allocr * alloc, struct ggml_tensor * tensor);
+GGML_API size_t ggml_allocr_alloc_graph(struct ggml_allocr * alloc, struct ggml_cgraph * graph);
+GGML_API size_t ggml_allocr_max_size   (struct ggml_allocr * alloc);
+
+GGML_API size_t ggml_allocr_alloc_graph_n(
+                    struct ggml_allocr * alloc,
+                    struct ggml_cgraph ** graphs, int n_graphs,
+                    struct ggml_tensor *** inputs, struct ggml_tensor *** outputs);
+
+#ifdef  __cplusplus
+}
+#endif

runner/ggml-backend.c

@@ -0,0 +1,411 @@
+/**
+ * llama.cpp - git 465219b9143ac01db0990bbcb0a081ef72ec2008
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023 Georgi Gerganov
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "ggml-backend.h"
+#include "ggml-alloc.h"
+
+#include <assert.h>
+#include <stdarg.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#define UNUSED GGML_UNUSED
+
+#define MAX(a, b) ((a) > (b) ? (a) : (b))
+
+// backend buffer
+
+ggml_backend_buffer_t ggml_backend_buffer_init(
+        struct ggml_backend                  * backend,
+        struct ggml_backend_buffer_i           iface,
+               ggml_backend_buffer_context_t   context,
+               size_t                          size) {
+    ggml_backend_buffer_t buffer = malloc(sizeof(struct ggml_backend_buffer));
+
+    GGML_ASSERT(iface.get_base != NULL);
+
+    (*buffer) = (struct ggml_backend_buffer) {
+        /* .interface = */ iface,
+        /* .backend   = */ backend,
+        /* .context   = */ context,
+        /* .size      = */ size,
+    };
+
+    return buffer;
+}
+
+void ggml_backend_buffer_free(ggml_backend_buffer_t buffer) {
+    if (buffer->iface.free_buffer != NULL) {
+        buffer->iface.free_buffer(buffer);
+    }
+    free(buffer);
+}
+
+size_t ggml_backend_buffer_get_alignment(ggml_backend_buffer_t buffer) {
+    return ggml_backend_get_alignment(buffer->backend);
+}
+
+void * ggml_backend_buffer_get_base(ggml_backend_buffer_t buffer) {
+    return buffer->iface.get_base(buffer);
+}
+
+size_t ggml_backend_buffer_get_size(ggml_backend_buffer_t buffer) {
+    return buffer->size;
+}
+
+size_t ggml_backend_buffer_get_alloc_size(ggml_backend_buffer_t buffer, struct ggml_tensor * tensor) {
+    if (buffer->iface.get_alloc_size) {
+        return buffer->iface.get_alloc_size(buffer, tensor);
+    }
+    return ggml_nbytes(tensor);
+}
+
+void ggml_backend_buffer_init_tensor(ggml_backend_buffer_t buffer, struct ggml_tensor * tensor) {
+    if (buffer->iface.init_tensor) {
+        buffer->iface.init_tensor(buffer, tensor);
+    }
+}
+
+void ggml_backend_buffer_free_tensor(ggml_backend_buffer_t buffer, struct ggml_tensor * tensor) {
+    if (buffer->iface.free_tensor) {
+        buffer->iface.free_tensor(buffer, tensor);
+    }
+}
+
+// backend
+
+ggml_backend_t ggml_get_backend(const struct ggml_tensor * tensor) {
+    return tensor->buffer->backend;
+}
+
+const char * ggml_backend_name(ggml_backend_t backend) {
+    return backend->iface.get_name(backend);
+}
+
+void ggml_backend_free(ggml_backend_t backend) {
+    backend->iface.free(backend);
+}
+
+ggml_backend_buffer_t ggml_backend_alloc_buffer(ggml_backend_t backend, size_t size) {
+    return backend->iface.alloc_buffer(backend, size);
+}
+
+size_t ggml_backend_get_alignment(ggml_backend_t backend) {
+    return backend->iface.get_alignment(backend);
+}
+
+void ggml_backend_tensor_set_async(struct ggml_tensor * tensor, const void * data, size_t offset, size_t size) {
+    ggml_get_backend(tensor)->iface.set_tensor_async(ggml_get_backend(tensor), tensor, data, offset, size);
+}
+
+void ggml_backend_tensor_get_async(const struct ggml_tensor * tensor, void * data, size_t offset, size_t size) {
+    ggml_get_backend(tensor)->iface.get_tensor_async(ggml_get_backend(tensor), tensor, data, offset, size);
+}
+
+void ggml_backend_tensor_set(struct ggml_tensor * tensor, const void * data, size_t offset, size_t size) {
+    ggml_get_backend(tensor)->iface.set_tensor_async(ggml_get_backend(tensor), tensor, data, offset, size);
+    ggml_get_backend(tensor)->iface.synchronize(ggml_get_backend(tensor));
+}
+
+void ggml_backend_tensor_get(const struct ggml_tensor * tensor, void * data, size_t offset, size_t size) {
+    ggml_get_backend(tensor)->iface.get_tensor_async(ggml_get_backend(tensor), tensor, data, offset, size);
+    ggml_get_backend(tensor)->iface.synchronize(ggml_get_backend(tensor));
+}
+
+void ggml_backend_synchronize(ggml_backend_t backend) {
+    backend->iface.synchronize(backend);
+}
+
+ggml_backend_graph_plan_t ggml_backend_graph_plan_create(ggml_backend_t backend, struct ggml_cgraph * cgraph) {
+    return backend->iface.graph_plan_create(backend, cgraph);
+}
+
+void ggml_backend_graph_plan_free(ggml_backend_t backend, ggml_backend_graph_plan_t plan) {
+    backend->iface.graph_plan_free(backend, plan);
+}
+
+void ggml_backend_graph_plan_compute(ggml_backend_t backend, ggml_backend_graph_plan_t plan) {
+    backend->iface.graph_plan_compute(backend, plan);
+}
+
+void ggml_backend_graph_compute(ggml_backend_t backend, struct ggml_cgraph * cgraph) {
+    backend->iface.graph_compute(backend, cgraph);
+}
+
+bool ggml_backend_supports_op(ggml_backend_t backend, const struct ggml_tensor * op) {
+    return backend->iface.supports_op(backend, op);
+}
+
+// backend copy
+
+static bool ggml_are_same_layout(const struct ggml_tensor * a, const struct ggml_tensor * b) {
+    if (a->type != b->type) {
+        return false;
+    }
+    for (int i = 0; i < GGML_MAX_DIMS; i++) {
+        if (a->ne[i] != b->ne[i]) {
+            return false;
+        }
+        if (a->nb[i] != b->nb[i]) {
+            return false;
+        }
+    }
+    return true;
+}
+
+void ggml_backend_tensor_copy(struct ggml_tensor * src, struct ggml_tensor * dst) {
+    //printf("src: %s ne: [%d %d %d %d] nb: [%d %d %d %d]\n", src->name, (int)src->ne[0], (int)src->ne[1], (int)src->ne[2], (int)src->ne[3], (int)src->nb[0], (int)src->nb[1], (int)src->nb[2], (int)src->nb[3]);
+    //printf("dst: %s ne: [%d %d %d %d] nb: [%d %d %d %d]\n", dst->name, (int)dst->ne[0], (int)dst->ne[1], (int)dst->ne[2], (int)dst->ne[3], (int)dst->nb[0], (int)dst->nb[1], (int)dst->nb[2], (int)dst->nb[3]);
+    GGML_ASSERT(ggml_are_same_layout(src, dst) && "cannot copy tensors with different layouts");
+
+    // printf("cpy tensor %s from %s to %s (%lu bytes)\n", src->name, ggml_backend_name(src->backend), ggml_backend_name(dst->backend), ggml_nbytes(src));
+
+    if (src == dst) {
+        return;
+    }
+
+    // TODO: allow backends to support copy to/from same backend
+
+    if (ggml_get_backend(dst)->iface.cpy_tensor_from != NULL) {
+        ggml_get_backend(dst)->iface.cpy_tensor_from(ggml_get_backend(dst)->context, src, dst);
+    } else if (ggml_get_backend(src)->iface.cpy_tensor_to != NULL) {
+        ggml_get_backend(src)->iface.cpy_tensor_to(ggml_get_backend(src)->context, src, dst);
+    } else {
+        // shouldn't be hit when copying from/to CPU
+        #ifndef NDEBUG
+        fprintf(stderr, "ggml_backend_tensor_copy: neither cpy_tensor_from nor cpy_tensor_to are implemented for backends %s and %s, falling back to get/set\n", ggml_backend_name(src->buffer->backend), ggml_backend_name(dst->buffer->backend));
+        #endif
+        size_t nbytes = ggml_nbytes(src);
+        void * data = malloc(nbytes);
+        ggml_backend_tensor_get(src, data, 0, nbytes);
+        ggml_backend_tensor_set(dst, data, 0, nbytes);
+        free(data);
+    }
+}
+
+// backend CPU
+
+struct ggml_backend_cpu_context {
+    int n_threads;
+    void * work_data;
+    size_t work_size;
+};
+
+static const char * ggml_backend_cpu_name(ggml_backend_t backend) {
+    return "CPU";
+
+    UNUSED(backend);
+}
+
+static void ggml_backend_cpu_free(ggml_backend_t backend) {
+    struct ggml_backend_cpu_context * cpu_ctx = (struct ggml_backend_cpu_context *)backend->context;
+    free(cpu_ctx->work_data);
+    free(cpu_ctx);
+    free(backend);
+}
+
+static void * ggml_backend_cpu_buffer_get_base(ggml_backend_buffer_t buffer) {
+    return (void *)buffer->context;
+}
+
+static void ggml_backend_cpu_buffer_free_buffer(ggml_backend_buffer_t buffer) {
+    free(buffer->context);
+    UNUSED(buffer);
+}
+
+static struct ggml_backend_buffer_i cpu_backend_buffer_i = {
+    /* .free_buffer    = */ ggml_backend_cpu_buffer_free_buffer,
+    /* .get_base       = */ ggml_backend_cpu_buffer_get_base,
+    /* .get_alloc_size = */ NULL, // defaults to ggml_nbytes
+    /* .init_tensor    = */ NULL, // no initialization required
+    /* .free_tensor    = */ NULL, // no cleanup required
+};
+
+// for buffers from ptr, free is not called
+static struct ggml_backend_buffer_i cpu_backend_buffer_i_from_ptr = {
+    /* .free_buffer    = */ NULL, // ptr is not owned by the buffer, so it does not need to be freed
+    /* .get_base       = */ ggml_backend_cpu_buffer_get_base,
+    /* .get_alloc_size = */ NULL, // defaults to ggml_nbytes
+    /* .init_tensor    = */ NULL,
+    /* .free_tensor    = */ NULL,
+};
+
+static const size_t TENSOR_ALIGNMENT = 64; // should be enough for AVX 512
+
+static ggml_backend_buffer_t ggml_backend_cpu_alloc_buffer(ggml_backend_t backend, size_t size) {
+    size += TENSOR_ALIGNMENT;   // malloc may return an address that is not aligned
+    void * data = malloc(size); // TODO: maybe use GGML_ALIGNED_MALLOC?
+
+    return ggml_backend_buffer_init(backend, cpu_backend_buffer_i, data, size);
+}
+
+static size_t ggml_backend_cpu_get_alignment(ggml_backend_t backend) {
+    return TENSOR_ALIGNMENT;
+    UNUSED(backend);
+}
+
+static void ggml_backend_cpu_set_tensor_async(ggml_backend_t backend, struct ggml_tensor * tensor, const void * data, size_t offset, size_t size) {
+    GGML_ASSERT(offset + size <= ggml_nbytes(tensor) && "tensor write out of bounds");
+    GGML_ASSERT(tensor->data != NULL && "tensor not allocated");
+
+    memcpy((char *)tensor->data + offset, data, size);
+
+    UNUSED(backend);
+}
+
+static void ggml_backend_cpu_get_tensor_async(ggml_backend_t backend, const struct ggml_tensor * tensor, void * data, size_t offset, size_t size) {
+    GGML_ASSERT(offset + size <= ggml_nbytes(tensor) && "tensor read out of bounds");
+    GGML_ASSERT(tensor->data != NULL && "tensor not allocated");
+
+    memcpy(data, (const char *)tensor->data + offset, size);
+
+    UNUSED(backend);
+}
+
+static void ggml_backend_cpu_synchronize(ggml_backend_t backend) {
+    UNUSED(backend);
+}
+
+static void ggml_backend_cpu_cpy_tensor_from(ggml_backend_t backend, struct ggml_tensor * src, struct ggml_tensor * dst) {
+    ggml_backend_tensor_get(src, dst->data, 0, ggml_nbytes(src));
+
+    UNUSED(backend);
+}
+
+static void ggml_backend_cpu_cpy_tensor_to(ggml_backend_t backend, struct ggml_tensor * src, struct ggml_tensor * dst) {
+    // for a backend such as CUDA that can queue async calls, it is ok to do this asynchronously, but it may not be the case for other backends
+    ggml_backend_tensor_set_async(dst, src->data, 0, ggml_nbytes(src));
+
+    UNUSED(backend);
+}
+
+struct ggml_backend_plan_cpu {
+    struct ggml_cplan cplan;
+    struct ggml_cgraph cgraph;
+};
+
+static ggml_backend_graph_plan_t ggml_backend_cpu_graph_plan_create(ggml_backend_t backend, struct ggml_cgraph * cgraph) {
+    struct ggml_backend_cpu_context * cpu_ctx = (struct ggml_backend_cpu_context *)backend->context;
+
+    struct ggml_backend_plan_cpu * cpu_plan = malloc(sizeof(struct ggml_backend_plan_cpu));
+
+    cpu_plan->cplan = ggml_graph_plan(cgraph, cpu_ctx->n_threads);
+    cpu_plan->cgraph = *cgraph;
+
+    if (cpu_plan->cplan.work_size > 0) {
+        cpu_plan->cplan.work_data = malloc(cpu_plan->cplan.work_size);
+    }
+
+    return cpu_plan;
+}
+
+static void ggml_backend_cpu_graph_plan_free(ggml_backend_t backend, ggml_backend_graph_plan_t plan) {
+    struct ggml_backend_plan_cpu * cpu_plan = (struct ggml_backend_plan_cpu *)plan;
+
+    free(cpu_plan->cplan.work_data);
+    free(cpu_plan);
+
+    UNUSED(backend);
+}
+
+static void ggml_backend_cpu_graph_plan_compute(ggml_backend_t backend, ggml_backend_graph_plan_t plan) {
+    struct ggml_backend_plan_cpu * cpu_plan = (struct ggml_backend_plan_cpu *)plan;
+
+    ggml_graph_compute(&cpu_plan->cgraph, &cpu_plan->cplan);
+
+    UNUSED(backend);
+}
+
+static void ggml_backend_cpu_graph_compute(ggml_backend_t backend, struct ggml_cgraph * cgraph) {
+    struct ggml_backend_cpu_context * cpu_ctx = (struct ggml_backend_cpu_context *)backend->context;
+
+    struct ggml_cplan cplan = ggml_graph_plan(cgraph, cpu_ctx->n_threads);
+
+    if (cpu_ctx->work_size < cplan.work_size) {
+        // TODO: may be faster to free and use malloc to avoid the copy
+        cpu_ctx->work_data = realloc(cpu_ctx->work_data, cplan.work_size);
+        cpu_ctx->work_size = cplan.work_size;
+    }
+
+    cplan.work_data = cpu_ctx->work_data;
+
+    ggml_graph_compute(cgraph, &cplan);
+}
+
+static bool ggml_backend_cpu_supports_op(ggml_backend_t backend, const struct ggml_tensor * op) {
+    return true;
+    UNUSED(backend);
+    UNUSED(op);
+}
+
+static struct ggml_backend_i cpu_backend_i = {
+    /* .get_name            = */ ggml_backend_cpu_name,
+    /* .free                = */ ggml_backend_cpu_free,
+    /* .alloc_buffer        = */ ggml_backend_cpu_alloc_buffer,
+    /* .get_alignment       = */ ggml_backend_cpu_get_alignment,
+    /* .set_tensor_async    = */ ggml_backend_cpu_set_tensor_async,
+    /* .get_tensor_async    = */ ggml_backend_cpu_get_tensor_async,
+    /* .synchronize         = */ ggml_backend_cpu_synchronize,
+    /* .cpy_tensor_from     = */ ggml_backend_cpu_cpy_tensor_from,
+    /* .cpy_tensor_to       = */ ggml_backend_cpu_cpy_tensor_to,
+    /* .graph_plan_create   = */ ggml_backend_cpu_graph_plan_create,
+    /* .graph_plan_free     = */ ggml_backend_cpu_graph_plan_free,
+    /* .graph_plan_compute  = */ ggml_backend_cpu_graph_plan_compute,
+    /* .graph_compute       = */ ggml_backend_cpu_graph_compute,
+    /* .supports_op         = */ ggml_backend_cpu_supports_op,
+};
+
+ggml_backend_t ggml_backend_cpu_init(void) {
+    struct ggml_backend_cpu_context * ctx = malloc(sizeof(struct ggml_backend_cpu_context));
+
+    ctx->n_threads = GGML_DEFAULT_N_THREADS;
+    ctx->work_data = NULL;
+    ctx->work_size = 0;
+
+    ggml_backend_t cpu_backend = malloc(sizeof(struct ggml_backend));
+
+    *cpu_backend = (struct ggml_backend) {
+        /* .interface = */ cpu_backend_i,
+        /* .context   = */ ctx
+    };
+    return cpu_backend;
+}
+
+bool ggml_backend_is_cpu(ggml_backend_t backend) {
+    return backend->iface.get_name == ggml_backend_cpu_name;
+}
+
+void ggml_backend_cpu_set_n_threads(ggml_backend_t backend_cpu, int n_threads) {
+    GGML_ASSERT(ggml_backend_is_cpu(backend_cpu));
+
+    struct ggml_backend_cpu_context * ctx = (struct ggml_backend_cpu_context *)backend_cpu->context;
+    ctx->n_threads = n_threads;
+}
+
+ggml_backend_buffer_t ggml_backend_cpu_buffer_from_ptr(ggml_backend_t backend_cpu, void * ptr, size_t size) {
+    return ggml_backend_buffer_init(backend_cpu, cpu_backend_buffer_i_from_ptr, ptr, size);
+}

runner/ggml-backend.h

@@ -0,0 +1,169 @@
+/**
+ * llama.cpp - git 465219b9143ac01db0990bbcb0a081ef72ec2008
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023 Georgi Gerganov
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#pragma once
+
+#include "ggml.h"
+
+#ifdef  __cplusplus
+extern "C" {
+#endif
+    struct ggml_backend;
+    struct ggml_backend_buffer;
+
+    // type-erased backend-specific types / wrappers
+    typedef void * ggml_backend_context_t;
+    typedef void * ggml_backend_graph_plan_t;
+    typedef void * ggml_backend_buffer_context_t;
+
+    // avoid accessing internals of these types
+    typedef struct ggml_backend        * ggml_backend_t;
+    typedef struct ggml_backend_buffer * ggml_backend_buffer_t;
+
+    //
+    // backend buffer
+    //
+
+    struct ggml_backend_buffer_i {
+        void   (*free_buffer)   (ggml_backend_buffer_t buffer);
+        void * (*get_base)      (ggml_backend_buffer_t buffer); // get base pointer
+        size_t (*get_alloc_size)(ggml_backend_buffer_t buffer, struct ggml_tensor * tensor); // pre-allocation callback
+        void   (*init_tensor)   (ggml_backend_buffer_t buffer, struct ggml_tensor * tensor); // post-allocation callback
+        void   (*free_tensor)   (ggml_backend_buffer_t buffer, struct ggml_tensor * tensor); // pre-free callback
+    };
+
+    // TODO: hide behind API
+    struct ggml_backend_buffer {
+        struct ggml_backend_buffer_i iface;
+
+        ggml_backend_t                backend;
+        ggml_backend_buffer_context_t context;
+
+        size_t size;
+    };
+
+    // backend buffer functions
+    GGML_API ggml_backend_buffer_t ggml_backend_buffer_init(
+            struct ggml_backend                  * backend,
+            struct ggml_backend_buffer_i           iface,
+                   ggml_backend_buffer_context_t   context,
+                   size_t                          size);
+
+    GGML_API void   ggml_backend_buffer_free          (ggml_backend_buffer_t buffer);
+    GGML_API size_t ggml_backend_buffer_get_alignment (ggml_backend_buffer_t buffer);
+    GGML_API void * ggml_backend_buffer_get_base      (ggml_backend_buffer_t buffer);
+    GGML_API size_t ggml_backend_buffer_get_size      (ggml_backend_buffer_t buffer);
+    GGML_API size_t ggml_backend_buffer_get_alloc_size(ggml_backend_buffer_t buffer, struct ggml_tensor * tensor);
+    GGML_API void   ggml_backend_buffer_init_tensor   (ggml_backend_buffer_t buffer, struct ggml_tensor * tensor);
+    GGML_API void   ggml_backend_buffer_free_tensor   (ggml_backend_buffer_t buffer, struct ggml_tensor * tensor);
+
+    //
+    // backend
+    //
+
+    struct ggml_backend_i {
+        const char * (*get_name)(ggml_backend_t backend);
+
+        void (*free)(ggml_backend_t backend);
+
+        // buffer allocation
+        ggml_backend_buffer_t (*alloc_buffer)(ggml_backend_t backend, size_t size);
+
+        // get buffer alignment
+        size_t (*get_alignment)(ggml_backend_t backend);
+
+        // tensor data access
+        // these functions can be asynchronous, helper functions are provided for synchronous access that automatically call synchronize
+        void (*set_tensor_async)(ggml_backend_t backend,       struct ggml_tensor * tensor, const void * data, size_t offset, size_t size);
+        void (*get_tensor_async)(ggml_backend_t backend, const struct ggml_tensor * tensor,       void * data, size_t offset, size_t size);
+        void (*synchronize)     (ggml_backend_t backend);
+
+        // (optional) copy tensor between different backends, allow for single-copy tranfers
+        void (*cpy_tensor_from)(ggml_backend_t backend, struct ggml_tensor * src, struct ggml_tensor * dst);
+        void (*cpy_tensor_to)  (ggml_backend_t backend, struct ggml_tensor * src, struct ggml_tensor * dst);
+
+        // compute graph with a plan
+        ggml_backend_graph_plan_t (*graph_plan_create) (ggml_backend_t backend, struct ggml_cgraph * cgraph);
+        void                      (*graph_plan_free)   (ggml_backend_t backend, ggml_backend_graph_plan_t plan);
+        void                      (*graph_plan_compute)(ggml_backend_t backend, ggml_backend_graph_plan_t plan);
+
+        // compute graph without a plan
+        void (*graph_compute)(ggml_backend_t backend, struct ggml_cgraph * cgraph);
+
+        // check if the backend supports an operation
+        bool (*supports_op)(ggml_backend_t backend, const struct ggml_tensor * op);
+    };
+
+    // TODO: hide behind API
+    struct ggml_backend {
+        struct ggml_backend_i iface;
+
+        ggml_backend_context_t context;
+    };
+
+    // backend helper functions
+    GGML_API ggml_backend_t ggml_get_backend(const struct ggml_tensor * tensor);
+
+    GGML_API const char * ggml_backend_name(ggml_backend_t backend);
+    GGML_API void         ggml_backend_free(ggml_backend_t backend);
+
+    GGML_API ggml_backend_buffer_t ggml_backend_alloc_buffer(ggml_backend_t backend, size_t size);
+
+    GGML_API size_t ggml_backend_get_alignment(ggml_backend_t backend);
+
+    GGML_API void ggml_backend_tensor_set_async(      struct ggml_tensor * tensor, const void * data, size_t offset, size_t size);
+    GGML_API void ggml_backend_tensor_get_async(const struct ggml_tensor * tensor,       void * data, size_t offset, size_t size);
+
+    GGML_API void ggml_backend_tensor_set(      struct ggml_tensor * tensor, const void * data, size_t offset, size_t size);
+    GGML_API void ggml_backend_tensor_get(const struct ggml_tensor * tensor,       void * data, size_t offset, size_t size);
+
+    GGML_API void ggml_backend_synchronize(ggml_backend_t backend);
+
+    GGML_API ggml_backend_graph_plan_t ggml_backend_graph_plan_create (ggml_backend_t backend, struct ggml_cgraph * cgraph);
+
+    GGML_API void ggml_backend_graph_plan_free   (ggml_backend_t backend, ggml_backend_graph_plan_t plan);
+    GGML_API void ggml_backend_graph_plan_compute(ggml_backend_t backend, ggml_backend_graph_plan_t plan);
+    GGML_API void ggml_backend_graph_compute     (ggml_backend_t backend, struct ggml_cgraph * cgraph);
+    GGML_API bool ggml_backend_supports_op       (ggml_backend_t backend, const struct ggml_tensor * op);
+
+    // tensor copy between different backends
+    GGML_API void ggml_backend_tensor_copy(struct ggml_tensor * src, struct ggml_tensor * dst);
+
+    //
+    // CPU backend
+    //
+
+    GGML_API ggml_backend_t ggml_backend_cpu_init(void);
+
+    GGML_API bool ggml_backend_is_cpu(ggml_backend_t backend);
+
+    GGML_API void ggml_backend_cpu_set_n_threads(ggml_backend_t backend_cpu, int n_threads);
+
+    GGML_API ggml_backend_buffer_t ggml_backend_cpu_buffer_from_ptr(ggml_backend_t backend_cpu, void * ptr, size_t size);
+
+#ifdef  __cplusplus
+}
+#endif

runner/ggml-cuda.h

@@ -0,0 +1,77 @@
+/**
+ * llama.cpp - git 465219b9143ac01db0990bbcb0a081ef72ec2008
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023 Georgi Gerganov
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#pragma once
+
+#include "ggml.h"
+#include "ggml-backend.h"
+
+#ifdef GGML_USE_HIPBLAS
+#define GGML_CUDA_NAME "ROCm"
+#define GGML_CUBLAS_NAME "hipBLAS"
+#else
+#define GGML_CUDA_NAME "CUDA"
+#define GGML_CUBLAS_NAME "cuBLAS"
+#endif
+
+#ifdef  __cplusplus
+extern "C" {
+#endif
+
+#define GGML_CUDA_MAX_DEVICES       16
+
+GGML_API void   ggml_init_cublas(void);
+GGML_API void * ggml_cuda_host_malloc(size_t size);
+GGML_API void   ggml_cuda_host_free(void * ptr);
+
+GGML_API bool   ggml_cuda_can_mul_mat(const struct ggml_tensor * src0, const struct ggml_tensor * src1, struct ggml_tensor * dst);
+GGML_API void   ggml_cuda_set_tensor_split(const float * tensor_split);
+GGML_API void   ggml_cuda_transform_tensor(void * data, struct ggml_tensor * tensor);
+GGML_API void   ggml_cuda_free_data(struct ggml_tensor * tensor);
+
+GGML_API void   ggml_cuda_assign_buffers(struct ggml_tensor * tensor);
+GGML_API void   ggml_cuda_assign_buffers_no_scratch(struct ggml_tensor * tensor);
+GGML_API void   ggml_cuda_assign_buffers_force_inplace(struct ggml_tensor * tensor);
+
+GGML_API void   ggml_cuda_assign_buffers_no_alloc(struct ggml_tensor * tensor);
+GGML_API void   ggml_cuda_assign_scratch_offset(struct ggml_tensor * tensor, size_t offset);
+GGML_API void   ggml_cuda_copy_to_device(struct ggml_tensor * tensor);
+
+GGML_API void   ggml_cuda_set_main_device(int main_device);
+GGML_API void   ggml_cuda_set_mul_mat_q(bool mul_mat_q);
+GGML_API void   ggml_cuda_set_scratch_size(size_t scratch_size);
+GGML_API void   ggml_cuda_free_scratch(void);
+GGML_API bool   ggml_cuda_compute_forward(struct ggml_compute_params * params, struct ggml_tensor * tensor);
+
+GGML_API int    ggml_cuda_get_device_count(void);
+GGML_API void   ggml_cuda_get_device_description(int device, char * description, size_t description_size);
+
+// backend API
+GGML_API ggml_backend_t ggml_backend_cuda_init(void); // TODO: take a list of devices to use
+
+#ifdef  __cplusplus
+}
+#endif

runner/ggml-metal.h

@@ -0,0 +1,134 @@
+//go:build darwin
+
+/**
+ * llama.cpp - git 465219b9143ac01db0990bbcb0a081ef72ec2008
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023 Georgi Gerganov
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+// An interface allowing to compute ggml_cgraph with Metal
+//
+// This is a fully functional interface that extends ggml with GPU support for Apple devices.
+// A similar interface can be created for other GPU backends (e.g. Vulkan, CUDA, OpenCL, etc.)
+//
+// How it works?
+//
+// As long as your program can create and evaluate a ggml_cgraph on the CPU, you can use this
+// interface to evaluate the same graph on the GPU. Instead of using ggml_graph_compute(), you
+// use ggml_metal_graph_compute() (or ggml_vulkan_graph_compute(), etc.)
+//
+// You only need to make sure that all memory buffers that you used during the graph creation
+// are mapped to the device memory with the ggml_metal_add_buffer() function. This mapping is
+// used during the graph evaluation to determine the arguments of the compute kernels.
+//
+// Synchronization between device and host memory (for example for input and output tensors)
+// is done with the ggml_metal_set_tensor() and ggml_metal_get_tensor() functions.
+//
+
+#pragma once
+
+#include "ggml.h"
+#include "ggml-backend.h"
+
+#include <stddef.h>
+#include <stdbool.h>
+
+// max memory buffers that can be mapped to the device
+#define GGML_METAL_MAX_BUFFERS 16
+#define GGML_METAL_MAX_COMMAND_BUFFERS 32
+
+struct ggml_tensor;
+struct ggml_cgraph;
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+//
+// internal API
+// temporary exposed to user-code
+//
+
+struct ggml_metal_context;
+
+void ggml_metal_log_set_callback(ggml_log_callback log_callback, void * user_data);
+
+// number of command buffers to use
+struct ggml_metal_context * ggml_metal_init(int n_cb);
+void ggml_metal_free(struct ggml_metal_context * ctx);
+
+void * ggml_metal_host_malloc(size_t n);
+void   ggml_metal_host_free  (void * data);
+
+// set the number of command buffers to use
+void ggml_metal_set_n_cb(struct ggml_metal_context * ctx, int n_cb);
+
+// creates a mapping between a host memory buffer and a device memory buffer
+// - make sure to map all buffers used in the graph before calling ggml_metal_graph_compute
+// - the mapping is used during computation to determine the arguments of the compute kernels
+// - you don't need to keep the host memory buffer allocated as it is never accessed by Metal
+// - max_size specifies the maximum size of a tensor and is used to create shared views such
+//   that it is guaranteed that the tensor will fit in at least one of the views
+//
+bool ggml_metal_add_buffer(
+        struct ggml_metal_context * ctx,
+                       const char * name,
+                             void * data,
+                           size_t   size,
+                           size_t   max_size);
+
+// set data from host memory into the device
+void ggml_metal_set_tensor(struct ggml_metal_context * ctx, struct ggml_tensor * t);
+
+// get data from the device into host memory
+void ggml_metal_get_tensor(struct ggml_metal_context * ctx, struct ggml_tensor * t);
+
+// try to find operations that can be run concurrently in the graph
+// you should run it again if the topology of your graph changes
+void ggml_metal_graph_find_concurrency(struct ggml_metal_context * ctx, struct ggml_cgraph * gf, bool check_mem);
+
+// if the graph has been optimized for concurrently dispatch, return length of the concur_list if optimized
+int ggml_metal_if_optimized(struct ggml_metal_context * ctx);
+
+// output the concur_list for ggml_alloc
+int * ggml_metal_get_concur_list(struct ggml_metal_context * ctx);
+
+// same as ggml_graph_compute but uses Metal
+// creates gf->n_threads command buffers in parallel
+void ggml_metal_graph_compute(struct ggml_metal_context * ctx, struct ggml_cgraph * gf);
+
+//
+// backend API
+// user-code should use only these functions
+//
+
+GGML_API ggml_backend_t ggml_backend_metal_init(void);
+
+GGML_API bool ggml_backend_is_metal(ggml_backend_t backend);
+
+GGML_API void ggml_backend_metal_set_n_cb(ggml_backend_t backend, int n_cb);
+
+#ifdef __cplusplus
+}
+#endif
+

runner/ggml-mpi.c

@@ -0,0 +1,244 @@
+//go:build mpi
+
+/**
+ * llama.cpp - git 465219b9143ac01db0990bbcb0a081ef72ec2008
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023 Georgi Gerganov
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "ggml-mpi.h"
+
+#include "ggml.h"
+
+#include <mpi.h>
+
+#include <stdio.h>
+#include <stdlib.h>
+
+#define MIN(a, b) ((a) < (b) ? (a) : (b))
+
+#define UNUSED GGML_UNUSED
+
+struct ggml_mpi_context {
+    int rank;
+    int size;
+};
+
+void ggml_mpi_backend_init(void) {
+    MPI_Init(NULL, NULL);
+}
+
+void ggml_mpi_backend_free(void) {
+    MPI_Finalize();
+}
+
+struct ggml_mpi_context * ggml_mpi_init(void) {
+    struct ggml_mpi_context * ctx = calloc(1, sizeof(struct ggml_mpi_context));
+
+    MPI_Comm_rank(MPI_COMM_WORLD, &ctx->rank);
+    MPI_Comm_size(MPI_COMM_WORLD, &ctx->size);
+
+    return ctx;
+}
+
+void ggml_mpi_free(struct ggml_mpi_context * ctx) {
+    free(ctx);
+}
+
+int ggml_mpi_rank(struct ggml_mpi_context * ctx) {
+    return ctx->rank;
+}
+
+void ggml_mpi_eval_init(
+        struct ggml_mpi_context * ctx_mpi,
+                            int * n_tokens,
+                            int * n_past,
+                            int * n_threads) {
+    UNUSED(ctx_mpi);
+
+    // synchronize the worker node parameters with the root node
+    MPI_Barrier(MPI_COMM_WORLD);
+
+    MPI_Bcast(n_tokens,  1, MPI_INT, 0, MPI_COMM_WORLD);
+    MPI_Bcast(n_past,    1, MPI_INT, 0, MPI_COMM_WORLD);
+    MPI_Bcast(n_threads, 1, MPI_INT, 0, MPI_COMM_WORLD);
+}
+
+static int ggml_graph_get_node_idx(struct ggml_cgraph * gf, const char * name) {
+    struct ggml_tensor * t = ggml_graph_get_tensor(gf, name);
+    if (t == NULL) {
+        fprintf(stderr, "%s: tensor %s not found\n", __func__, name);
+        return -1;
+    }
+
+    for (int i = 0; i < gf->n_nodes; i++) {
+        if (gf->nodes[i] == t) {
+            return i;
+        }
+    }
+
+    fprintf(stderr, "%s: tensor %s not found in graph (should not happen)\n", __func__, name);
+    return -1;
+}
+
+static void ggml_mpi_tensor_send(struct ggml_tensor * t, int mpi_rank_dst) {
+    MPI_Datatype mpi_type;
+
+    switch (t->type) {
+        case GGML_TYPE_I32: mpi_type = MPI_INT32_T; break;
+        case GGML_TYPE_F32: mpi_type = MPI_FLOAT;   break;
+        default: GGML_ASSERT(false && "not implemented");
+    }
+
+    const int retval = MPI_Send(t->data, ggml_nelements(t), mpi_type, mpi_rank_dst, 0, MPI_COMM_WORLD);
+    GGML_ASSERT(retval == MPI_SUCCESS);
+}
+
+static void ggml_mpi_tensor_recv(struct ggml_tensor * t, int mpi_rank_src) {
+    MPI_Datatype mpi_type;
+
+    switch (t->type) {
+        case GGML_TYPE_I32: mpi_type = MPI_INT32_T; break;
+        case GGML_TYPE_F32: mpi_type = MPI_FLOAT;   break;
+        default: GGML_ASSERT(false && "not implemented");
+    }
+
+    MPI_Status status; UNUSED(status);
+
+    const int retval = MPI_Recv(t->data, ggml_nelements(t), mpi_type, mpi_rank_src, MPI_ANY_TAG, MPI_COMM_WORLD, &status);
+    GGML_ASSERT(retval == MPI_SUCCESS);
+}
+
+// TODO: there are many improvements that can be done to this implementation
+void ggml_mpi_graph_compute_pre(
+        struct ggml_mpi_context * ctx_mpi,
+             struct ggml_cgraph * gf,
+                            int   n_layers) {
+    const int mpi_rank = ctx_mpi->rank;
+    const int mpi_size = ctx_mpi->size;
+
+    struct ggml_tensor * inp_tokens = ggml_graph_get_tensor(gf, "inp_tokens");
+    if (inp_tokens == NULL) {
+        fprintf(stderr, "%s: tensor 'inp_tokens' not found\n", __func__);
+        return;
+    }
+
+    struct ggml_tensor * inp0 = ggml_graph_get_tensor(gf, "layer_inp_0");
+    if (inp0 == NULL) {
+        fprintf(stderr, "%s: tensor 'inp0' not found\n", __func__);
+        return;
+    }
+
+    GGML_ASSERT(inp0 == gf->nodes[0]);
+
+    // distribute the compute graph into slices across the MPI nodes
+    //
+    // the main node (0) processes the last layers + the remainder of the compute graph
+    // and is responsible to pass the input tokens to the first node (1)
+    //
+    // node 1:   [(  0) * n_per_node, (  1) * n_per_node)
+    // node 2:   [(  1) * n_per_node, (  2) * n_per_node)
+    // ...
+    // node n-1: [(n-2) * n_per_node, (n-1) * n_per_node)
+    // node 0:   [(n-1) * n_per_node,            n_nodes)
+    //
+    if (mpi_rank > 0) {
+        if (mpi_rank == 1) {
+            // the first node (1) receives the input tokens from the main node (0)
+            ggml_mpi_tensor_recv(inp_tokens, 0);
+        } else {
+            // recv input data for each node into the "inp0" tensor (i.e. the first node in the compute graph)
+            ggml_mpi_tensor_recv(inp0, mpi_rank - 1);
+        }
+    } else if (mpi_size > 1) {
+        // node 0 sends the input tokens to node 1
+        ggml_mpi_tensor_send(inp_tokens, 1);
+
+        // recv the output data from the last node
+        ggml_mpi_tensor_recv(inp0, mpi_size - 1);
+    }
+
+    {
+        const int n_per_node = (n_layers + (mpi_size - 1)) / mpi_size;
+
+        const int mpi_idx = mpi_rank > 0 ? mpi_rank - 1 : mpi_size - 1;
+
+        const int il0 =               (mpi_idx + 0) * n_per_node;
+        const int il1 = MIN(n_layers, (mpi_idx + 1) * n_per_node);
+
+        char name_l0[GGML_MAX_NAME];
+        char name_l1[GGML_MAX_NAME];
+
+        snprintf(name_l0, sizeof(name_l0), "layer_inp_%d", il0);
+        snprintf(name_l1, sizeof(name_l1), "layer_inp_%d", il1);
+
+        const int idx_l0 =                ggml_graph_get_node_idx(gf, name_l0);
+        const int idx_l1 = mpi_rank > 0 ? ggml_graph_get_node_idx(gf, name_l1) + 1 : gf->n_nodes;
+
+        if (idx_l0 < 0 || idx_l1 < 0) {
+            fprintf(stderr, "%s: layer input nodes not found\n", __func__);
+            return;
+        }
+
+        // attach the input data to all nodes that need it
+        // TODO: not great - should be able to do this without modifying the compute graph (see next TODO below)
+        for (int i = idx_l0; i < idx_l1; i++) {
+            if (gf->nodes[i]->src[0] == gf->nodes[idx_l0]) {
+                gf->nodes[i]->src[0] =  inp0;
+            }
+            if (gf->nodes[i]->src[1] == gf->nodes[idx_l0]) {
+                gf->nodes[i]->src[1] =  inp0;
+            }
+        }
+
+        // TODO: instead of rearranging the nodes, we should be able to execute a subset of the compute graph
+        for (int i = 1; i < idx_l1 - idx_l0; i++) {
+            gf->nodes[i] = gf->nodes[idx_l0 + i];
+            gf->grads[i] = gf->grads[idx_l0 + i];
+        }
+
+        // the first node performs the "get_rows" operation, the rest of the nodes get the data from the previous node
+        if (mpi_idx != 0) {
+            gf->nodes[0]->op = GGML_OP_NONE;
+        }
+
+        gf->n_nodes = idx_l1 - idx_l0;
+
+        //fprintf(stderr, "%s: node %d: processing %d nodes [%d, %d)\n", __func__, mpi_rank, gf->n_nodes, il0, il1);
+    }
+}
+
+void ggml_mpi_graph_compute_post(
+        struct ggml_mpi_context * ctx_mpi,
+             struct ggml_cgraph * gf,
+                            int   n_layers) {
+    UNUSED(n_layers);
+
+    const int mpi_rank = ctx_mpi->rank;
+    const int mpi_size = ctx_mpi->size;
+
+    // send the output data to the next node
+    if (mpi_rank > 0) {
+        ggml_mpi_tensor_send(gf->nodes[gf->n_nodes - 1], (mpi_rank + 1) % mpi_size);
+    }
+}

runner/ggml-mpi.h

@@ -0,0 +1,67 @@
+//go:build mpi
+
+/**
+ * llama.cpp - git 465219b9143ac01db0990bbcb0a081ef72ec2008
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023 Georgi Gerganov
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#pragma once
+
+struct ggml_context;
+struct ggml_tensor;
+struct ggml_cgraph;
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+struct ggml_mpi_context;
+
+void ggml_mpi_backend_init(void);
+void ggml_mpi_backend_free(void);
+
+struct ggml_mpi_context * ggml_mpi_init(void);
+void ggml_mpi_free(struct ggml_mpi_context * ctx);
+
+int ggml_mpi_rank(struct ggml_mpi_context * ctx);
+
+void ggml_mpi_eval_init(
+        struct ggml_mpi_context * ctx_mpi,
+                            int * n_tokens,
+                            int * n_past,
+                            int * n_threads);
+
+void ggml_mpi_graph_compute_pre(
+        struct ggml_mpi_context * ctx_mpi,
+             struct ggml_cgraph * gf,
+                            int   n_layers);
+
+void ggml_mpi_graph_compute_post(
+        struct ggml_mpi_context * ctx_mpi,
+             struct ggml_cgraph * gf,
+                            int   n_layers);
+
+#ifdef __cplusplus
+}
+#endif

runner/ggml-opencl.h

@@ -0,0 +1,53 @@
+//go:build opencl
+
+/**
+ * llama.cpp - git 465219b9143ac01db0990bbcb0a081ef72ec2008
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023 Georgi Gerganov
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#pragma once
+
+#include "ggml.h"
+
+#ifdef  __cplusplus
+extern "C" {
+#endif
+
+void ggml_cl_init(void);
+
+void   ggml_cl_mul(const struct ggml_tensor * src0, const struct ggml_tensor * src1, struct ggml_tensor * dst);
+bool   ggml_cl_can_mul_mat(const struct ggml_tensor * src0, const struct ggml_tensor * src1, struct ggml_tensor * dst);
+size_t ggml_cl_mul_mat_get_wsize(const struct ggml_tensor * src0, const struct ggml_tensor * src1, struct ggml_tensor * dst);
+void   ggml_cl_mul_mat(const struct ggml_tensor * src0, const struct ggml_tensor * src1, struct ggml_tensor * dst, void * wdata, size_t wsize);
+
+void * ggml_cl_host_malloc(size_t size);
+void   ggml_cl_host_free(void * ptr);
+
+void ggml_cl_free_data(const struct ggml_tensor* tensor);
+
+void ggml_cl_transform_tensor(void * data, struct ggml_tensor * tensor);
+
+#ifdef  __cplusplus
+}
+#endif

runner/k_quants.h

@@ -0,0 +1,191 @@
+/**
+ * llama.cpp - git 465219b9143ac01db0990bbcb0a081ef72ec2008
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023 Georgi Gerganov
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#pragma once
+
+#include "ggml.h"
+
+#include <stdint.h>
+#include <assert.h>
+#include <stddef.h>
+
+// Super-block size
+#ifdef GGML_QKK_64
+#define QK_K 64
+#define K_SCALE_SIZE 4
+#else
+#define QK_K 256
+#define K_SCALE_SIZE 12
+#endif
+
+#ifndef static_assert
+#if defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 201100L)
+#define static_assert(cond, msg) _Static_assert(cond, msg)
+#else
+#define static_assert(cond, msg) struct global_scope_noop_trick
+#endif
+#endif
+
+//
+// Super-block quantization structures
+//
+
+// 2-bit quantization
+// weight is represented as x = a * q + b
+// 16 blocks of 16 elements each
+// Effectively 2.5625 bits per weight
+typedef struct {
+    uint8_t scales[QK_K/16]; // scales and mins, quantized with 4 bits
+    uint8_t qs[QK_K/4];      // quants
+    ggml_fp16_t d;           // super-block scale for quantized scales
+    ggml_fp16_t dmin;        // super-block scale for quantized mins
+} block_q2_K;
+static_assert(sizeof(block_q2_K) == 2*sizeof(ggml_fp16_t) + QK_K/16 + QK_K/4, "wrong q2_K block size/padding");
+
+// 3-bit quantization
+// weight is represented as x = a * q
+// 16 blocks of 16 elements each
+// Effectively 3.4375 bits per weight
+#ifdef GGML_QKK_64
+typedef struct {
+    uint8_t hmask[QK_K/8];     // quants - high bit
+    uint8_t qs[QK_K/4];        // quants - low 2 bits
+    uint8_t scales[2];
+    ggml_fp16_t d;             // super-block scale
+} block_q3_K;
+static_assert(sizeof(block_q3_K) == sizeof(ggml_fp16_t) + QK_K / 4 + QK_K / 8 + 2, "wrong q3_K block size/padding");
+#else
+typedef struct {
+    uint8_t hmask[QK_K/8];     // quants - high bit
+    uint8_t qs[QK_K/4];        // quants - low 2 bits
+    uint8_t scales[12];        // scales, quantized with 6 bits
+    ggml_fp16_t d;             // super-block scale
+} block_q3_K;
+static_assert(sizeof(block_q3_K) == sizeof(ggml_fp16_t) + QK_K / 4 + QK_K / 8 + 12, "wrong q3_K block size/padding");
+#endif
+
+// 4-bit quantization
+// 8 blocks of 32 elements each
+// weight is represented as x = a * q + b
+// Effectively 4.5 bits per weight
+#ifdef GGML_QKK_64
+typedef struct {
+    ggml_fp16_t d[2];          // super-block scales/mins
+    uint8_t scales[2];         // 4-bit block scales/mins
+    uint8_t qs[QK_K/2];        // 4--bit quants
+} block_q4_K;
+static_assert(sizeof(block_q4_K) == 2*sizeof(ggml_fp16_t) + QK_K/2 + 2, "wrong q4_K block size/padding");
+#else
+typedef struct {
+    ggml_fp16_t d;             // super-block scale for quantized scales
+    ggml_fp16_t dmin;          // super-block scale for quantized mins
+    uint8_t scales[K_SCALE_SIZE]; // scales and mins, quantized with 6 bits
+    uint8_t qs[QK_K/2];        // 4--bit quants
+} block_q4_K;
+static_assert(sizeof(block_q4_K) == 2*sizeof(ggml_fp16_t) + K_SCALE_SIZE + QK_K/2, "wrong q4_K block size/padding");
+#endif
+
+// 5-bit quantization
+// 8 blocks of 32 elements each
+// weight is represented as x = a * q + b
+// Effectively 5.5 bits per weight
+#ifdef GGML_QKK_64
+typedef struct {
+    ggml_fp16_t d;               // super-block scale
+    int8_t  scales[QK_K/16];     // 8-bit block scales
+    uint8_t qh[QK_K/8];          // quants, high bit
+    uint8_t qs[QK_K/2];          // quants, low 4 bits
+} block_q5_K;
+static_assert(sizeof(block_q5_K) == sizeof(ggml_fp16_t) + QK_K/2 + QK_K/8 + QK_K/16, "wrong q5_K block size/padding");
+#else
+typedef struct {
+    ggml_fp16_t d;               // super-block scale for quantized scales
+    ggml_fp16_t dmin;            // super-block scale for quantized mins
+    uint8_t scales[K_SCALE_SIZE];   // scales and mins, quantized with 6 bits
+    uint8_t qh[QK_K/8];          // quants, high bit
+    uint8_t qs[QK_K/2];          // quants, low 4 bits
+} block_q5_K;
+static_assert(sizeof(block_q5_K) == 2*sizeof(ggml_fp16_t) + K_SCALE_SIZE + QK_K/2 + QK_K/8, "wrong q5_K block size/padding");
+#endif
+
+// 6-bit quantization
+// weight is represented as x = a * q
+// 16 blocks of 16 elements each
+// Effectively 6.5625 bits per weight
+typedef struct {
+    uint8_t ql[QK_K/2];      // quants, lower 4 bits
+    uint8_t qh[QK_K/4];      // quants, upper 2 bits
+    int8_t  scales[QK_K/16]; // scales, quantized with 8 bits
+    ggml_fp16_t d;           // super-block scale
+} block_q6_K;
+static_assert(sizeof(block_q6_K) == sizeof(ggml_fp16_t) + QK_K / 16 + 3*QK_K/4, "wrong q6_K block size/padding");
+
+// This is only used for intermediate quantization and dot products
+typedef struct {
+    float   d;              // delta
+    int8_t  qs[QK_K];       // quants
+    int16_t bsums[QK_K/16]; // sum of quants in groups of 16
+} block_q8_K;
+static_assert(sizeof(block_q8_K) == sizeof(float) + QK_K + QK_K/16*sizeof(int16_t), "wrong q8_K block size/padding");
+
+
+// Quantization
+void quantize_row_q2_K_reference(const float * restrict x, block_q2_K * restrict y, int k);
+void quantize_row_q3_K_reference(const float * restrict x, block_q3_K * restrict y, int k);
+void quantize_row_q4_K_reference(const float * restrict x, block_q4_K * restrict y, int k);
+void quantize_row_q5_K_reference(const float * restrict x, block_q5_K * restrict y, int k);
+void quantize_row_q6_K_reference(const float * restrict x, block_q6_K * restrict y, int k);
+void quantize_row_q8_K_reference(const float * restrict x, block_q8_K * restrict y, int k);
+
+void quantize_row_q2_K(const float * restrict x, void * restrict y, int k);
+void quantize_row_q3_K(const float * restrict x, void * restrict y, int k);
+void quantize_row_q4_K(const float * restrict x, void * restrict y, int k);
+void quantize_row_q5_K(const float * restrict x, void * restrict y, int k);
+void quantize_row_q6_K(const float * restrict x, void * restrict y, int k);
+void quantize_row_q8_K(const float * restrict x, void * restrict y, int k);
+
+// Dequantization
+void dequantize_row_q2_K(const block_q2_K * restrict x, float * restrict y, int k);
+void dequantize_row_q3_K(const block_q3_K * restrict x, float * restrict y, int k);
+void dequantize_row_q4_K(const block_q4_K * restrict x, float * restrict y, int k);
+void dequantize_row_q5_K(const block_q5_K * restrict x, float * restrict y, int k);
+void dequantize_row_q6_K(const block_q6_K * restrict x, float * restrict y, int k);
+void dequantize_row_q8_K(const block_q8_K * restrict x, float * restrict y, int k);
+
+// Dot product
+void ggml_vec_dot_q2_K_q8_K(int n, float * restrict s, const void * restrict vx, const void * restrict vy);
+void ggml_vec_dot_q3_K_q8_K(int n, float * restrict s, const void * restrict vx, const void * restrict vy);
+void ggml_vec_dot_q4_K_q8_K(int n, float * restrict s, const void * restrict vx, const void * restrict vy);
+void ggml_vec_dot_q5_K_q8_K(int n, float * restrict s, const void * restrict vx, const void * restrict vy);
+void ggml_vec_dot_q6_K_q8_K(int n, float * restrict s, const void * restrict vx, const void * restrict vy);
+
+// Quantization with histogram collection
+size_t ggml_quantize_q2_K(const float * src, void * dst, int n, int k, int64_t * hist);
+size_t ggml_quantize_q3_K(const float * src, void * dst, int n, int k, int64_t * hist);
+size_t ggml_quantize_q4_K(const float * src, void * dst, int n, int k, int64_t * hist);
+size_t ggml_quantize_q5_K(const float * src, void * dst, int n, int k, int64_t * hist);
+size_t ggml_quantize_q6_K(const float * src, void * dst, int n, int k, int64_t * hist);
+

runner/llama.h

@@ -0,0 +1,778 @@
+/**
+ * llama.cpp - git 465219b9143ac01db0990bbcb0a081ef72ec2008
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023 Georgi Gerganov
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef LLAMA_H
+#define LLAMA_H
+
+#include "ggml.h"
+#ifdef GGML_USE_CUBLAS
+#include "ggml-cuda.h"
+#define LLAMA_MAX_DEVICES GGML_CUDA_MAX_DEVICES
+#else
+#define LLAMA_MAX_DEVICES 1
+#endif // GGML_USE_CUBLAS
+#include <stddef.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <stdbool.h>
+
+#ifdef LLAMA_SHARED
+#    if defined(_WIN32) && !defined(__MINGW32__)
+#        ifdef LLAMA_BUILD
+#            define LLAMA_API __declspec(dllexport)
+#        else
+#            define LLAMA_API __declspec(dllimport)
+#        endif
+#    else
+#        define LLAMA_API __attribute__ ((visibility ("default")))
+#    endif
+#else
+#    define LLAMA_API
+#endif
+
+#ifdef __GNUC__
+#    define DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
+#elif defined(_MSC_VER)
+#    define DEPRECATED(func, hint) __declspec(deprecated(hint)) func
+#else
+#    define DEPRECATED(func, hint) func
+#endif
+
+#define LLAMA_DEFAULT_SEED 0xFFFFFFFF
+
+#define LLAMA_MAX_RNG_STATE (64*1024)
+
+#define LLAMA_FILE_MAGIC_GGSN 0x6767736eu // 'ggsn'
+
+#define LLAMA_SESSION_MAGIC   LLAMA_FILE_MAGIC_GGSN
+#define LLAMA_SESSION_VERSION 2
+
+#if defined(GGML_USE_CUBLAS) || defined(GGML_USE_CLBLAST) || defined(GGML_USE_METAL)
+// Defined when llama.cpp is compiled with support for offloading model layers to GPU.
+#define LLAMA_SUPPORTS_GPU_OFFLOAD
+#endif
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+    //
+    // C interface
+    //
+    // TODO: show sample usage
+    //
+
+    struct llama_model;
+    struct llama_context;
+
+    typedef int32_t llama_pos;
+    typedef int32_t llama_token;
+    typedef int32_t llama_seq_id;
+
+    enum llama_vocab_type {
+        LLAMA_VOCAB_TYPE_SPM = 0, // SentencePiece
+        LLAMA_VOCAB_TYPE_BPE = 1, // Byte Pair Encoding
+    };
+
+    enum llama_token_type {
+        LLAMA_TOKEN_TYPE_UNDEFINED    = 0,
+        LLAMA_TOKEN_TYPE_NORMAL       = 1,
+        LLAMA_TOKEN_TYPE_UNKNOWN      = 2,
+        LLAMA_TOKEN_TYPE_CONTROL      = 3,
+        LLAMA_TOKEN_TYPE_USER_DEFINED = 4,
+        LLAMA_TOKEN_TYPE_UNUSED       = 5,
+        LLAMA_TOKEN_TYPE_BYTE         = 6,
+    };
+
+    // model file types
+    enum llama_ftype {
+        LLAMA_FTYPE_ALL_F32              = 0,
+        LLAMA_FTYPE_MOSTLY_F16           = 1,  // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q4_0          = 2,  // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q4_1          = 3,  // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16 = 4,  // tok_embeddings.weight and output.weight are F16
+        // LLAMA_FTYPE_MOSTLY_Q4_2       = 5,  // support has been removed
+        // LLAMA_FTYPE_MOSTLY_Q4_3       = 6,  // support has been removed
+        LLAMA_FTYPE_MOSTLY_Q8_0          = 7,  // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q5_0          = 8,  // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q5_1          = 9,  // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q2_K          = 10, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q3_K_S        = 11, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q3_K_M        = 12, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q3_K_L        = 13, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q4_K_S        = 14, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q4_K_M        = 15, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q5_K_S        = 16, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q5_K_M        = 17, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q6_K          = 18, // except 1d tensors
+
+        LLAMA_FTYPE_GUESSED = 1024, // not specified in the model file
+    };
+
+    typedef struct llama_token_data {
+        llama_token id; // token id
+        float logit;    // log-odds of the token
+        float p;        // probability of the token
+    } llama_token_data;
+
+    typedef struct llama_token_data_array {
+        llama_token_data * data;
+        size_t size;
+        bool sorted;
+    } llama_token_data_array;
+
+    typedef void (*llama_progress_callback)(float progress, void *ctx);
+
+    // Input data for llama_decode
+    // A llama_batch object can contain input about one or many sequences
+    // The provided arrays (i.e. token, embd, pos, etc.) must have size of n_tokens
+    //
+    // - token  : the token ids of the input (used when embd is NULL)
+    // - embd   : token embeddings (i.e. float vector of size n_embd) (used when token is NULL)
+    // - pos    : the positions of the respective token in the sequence
+    // - seq_id : the sequence to which the respective token belongs
+    // - logits : if zero, the logits for the respective token will not be output
+    //
+    typedef struct llama_batch {
+        int32_t n_tokens;
+
+        llama_token  *  token;
+        float        *  embd;
+        llama_pos    *  pos;
+        int32_t      *  n_seq_id;
+        llama_seq_id ** seq_id;
+        int8_t       *  logits;
+
+        // NOTE: helpers for smooth API transition - can be deprecated in the future
+        //       for future-proof code, use the above fields instead and ignore everything below
+        //
+        // pos[i] = all_pos_0 + i*all_pos_1
+        //
+        llama_pos    all_pos_0;  // used if pos == NULL
+        llama_pos    all_pos_1;  // used if pos == NULL
+        llama_seq_id all_seq_id; // used if seq_id == NULL
+    } llama_batch;
+
+    struct llama_model_params {
+        int32_t n_gpu_layers; // number of layers to store in VRAM
+        int32_t main_gpu;     // the GPU that is used for scratch and small tensors
+        const float * tensor_split; // how to split layers across multiple GPUs (size: LLAMA_MAX_DEVICES)
+
+        // called with a progress value between 0 and 1, pass NULL to disable
+        llama_progress_callback progress_callback;
+        // context pointer passed to the progress callback
+        void * progress_callback_user_data;
+
+        // Keep the booleans together to avoid misalignment during copy-by-value.
+        bool vocab_only; // only load the vocabulary, no weights
+        bool use_mmap;   // use mmap if possible
+        bool use_mlock;  // force system to keep model in RAM
+    };
+
+    struct llama_context_params {
+        uint32_t seed;            // RNG seed, -1 for random
+        uint32_t n_ctx;           // text context, 0 = from model
+        uint32_t n_batch;         // prompt processing maximum batch size
+        uint32_t n_threads;       // number of threads to use for generation
+        uint32_t n_threads_batch; // number of threads to use for batch processing
+
+        // ref: https://github.com/ggerganov/llama.cpp/pull/2054
+        float rope_freq_base;  // RoPE base frequency, 0 = from model
+        float rope_freq_scale; // RoPE frequency scaling factor, 0 = from model
+
+        // Keep the booleans together to avoid misalignment during copy-by-value.
+        bool mul_mat_q;  // if true, use experimental mul_mat_q kernels
+        bool f16_kv;     // use fp16 for KV cache, fp32 otherwise
+        bool logits_all; // the llama_eval() call computes all logits, not just the last one
+        bool embedding;  // embedding mode only
+    };
+
+    // model quantization parameters
+    typedef struct llama_model_quantize_params {
+        int nthread;                 // number of threads to use for quantizing, if <=0 will use std::thread::hardware_concurrency()
+        enum llama_ftype ftype;      // quantize to this llama_ftype
+        bool allow_requantize;       // allow quantizing non-f32/f16 tensors
+        bool quantize_output_tensor; // quantize output.weight
+        bool only_copy;              // only copy tensors - ftype, allow_requantize and quantize_output_tensor are ignored
+    } llama_model_quantize_params;
+
+    // grammar types
+    struct llama_grammar;
+
+    // grammar element type
+    enum llama_gretype {
+        // end of rule definition
+        LLAMA_GRETYPE_END            = 0,
+
+        // start of alternate definition for rule
+        LLAMA_GRETYPE_ALT            = 1,
+
+        // non-terminal element: reference to rule
+        LLAMA_GRETYPE_RULE_REF       = 2,
+
+        // terminal element: character (code point)
+        LLAMA_GRETYPE_CHAR           = 3,
+
+        // inverse char(s) ([^a], [^a-b] [^abc])
+        LLAMA_GRETYPE_CHAR_NOT       = 4,
+
+        // modifies a preceding LLAMA_GRETYPE_CHAR or LLAMA_GRETYPE_CHAR_ALT to
+        // be an inclusive range ([a-z])
+        LLAMA_GRETYPE_CHAR_RNG_UPPER = 5,
+
+        // modifies a preceding LLAMA_GRETYPE_CHAR or
+        // LLAMA_GRETYPE_CHAR_RNG_UPPER to add an alternate char to match ([ab], [a-zA])
+        LLAMA_GRETYPE_CHAR_ALT       = 6,
+    };
+
+    typedef struct llama_grammar_element {
+        enum llama_gretype type;
+        uint32_t           value; // Unicode code point or rule ID
+    } llama_grammar_element;
+
+    // performance timing information
+    struct llama_timings {
+        double t_start_ms;
+        double t_end_ms;
+        double t_load_ms;
+        double t_sample_ms;
+        double t_p_eval_ms;
+        double t_eval_ms;
+
+        int32_t n_sample;
+        int32_t n_p_eval;
+        int32_t n_eval;
+    };
+
+    // Helpers for getting default parameters
+    LLAMA_API struct llama_model_params llama_model_default_params(void);
+    LLAMA_API struct llama_context_params llama_context_default_params(void);
+    LLAMA_API struct llama_model_quantize_params llama_model_quantize_default_params(void);
+
+    // Initialize the llama + ggml backend
+    // If numa is true, use NUMA optimizations
+    // Call once at the start of the program
+    LLAMA_API void llama_backend_init(bool numa);
+
+    // Call once at the end of the program - currently only used for MPI
+    LLAMA_API void llama_backend_free(void);
+
+    LLAMA_API struct llama_model * llama_load_model_from_file(
+                             const char * path_model,
+            struct llama_model_params     params);
+
+    LLAMA_API void llama_free_model(struct llama_model * model);
+
+    LLAMA_API struct llama_context * llama_new_context_with_model(
+                     struct llama_model * model,
+            struct llama_context_params   params);
+
+    // Frees all allocated memory
+    LLAMA_API void llama_free(struct llama_context * ctx);
+
+    LLAMA_API int64_t llama_time_us(void);
+
+    LLAMA_API int  llama_max_devices    (void);
+    LLAMA_API bool llama_mmap_supported (void);
+    LLAMA_API bool llama_mlock_supported(void);
+
+    LLAMA_API const struct llama_model * llama_get_model(const struct llama_context * ctx);
+
+    LLAMA_API int llama_n_ctx      (const struct llama_context * ctx);
+
+    LLAMA_API enum llama_vocab_type llama_vocab_type(const struct llama_model * model);
+
+    LLAMA_API int llama_n_vocab    (const struct llama_model * model);
+    LLAMA_API int llama_n_ctx_train(const struct llama_model * model);
+    LLAMA_API int llama_n_embd     (const struct llama_model * model);
+
+    // Get the model's RoPE frequency scaling factor
+    LLAMA_API float llama_rope_freq_scale_train(const struct llama_model * model);
+
+    // Get a string describing the model type
+    LLAMA_API int llama_model_desc(const struct llama_model * model, char * buf, size_t buf_size);
+
+    // Returns the total size of all the tensors in the model in bytes
+    LLAMA_API uint64_t llama_model_size(const struct llama_model * model);
+
+    // Returns the total number of parameters in the model
+    LLAMA_API uint64_t llama_model_n_params(const struct llama_model * model);
+
+    // Get a llama model tensor
+    LLAMA_API struct ggml_tensor * llama_get_model_tensor(struct llama_model * model, const char * name);
+
+    // Returns 0 on success
+    LLAMA_API int llama_model_quantize(
+            const char * fname_inp,
+            const char * fname_out,
+            const llama_model_quantize_params * params);
+
+    // Apply a LoRA adapter to a loaded model
+    // path_base_model is the path to a higher quality model to use as a base for
+    // the layers modified by the adapter. Can be NULL to use the current loaded model.
+    // The model needs to be reloaded before applying a new adapter, otherwise the adapter
+    // will be applied on top of the previous one
+    // Returns 0 on success
+    LLAMA_API DEPRECATED(int llama_apply_lora_from_file(
+            struct llama_context * ctx,
+                      const char * path_lora,
+                           float   scale,
+                      const char * path_base_model,
+                             int   n_threads),
+            "use llama_model_apply_lora_from_file instead");
+
+    LLAMA_API int llama_model_apply_lora_from_file(
+            const struct llama_model * model,
+                      const char * path_lora,
+                           float   scale,
+                      const char * path_base_model,
+                             int   n_threads);
+
+    //
+    // KV cache
+    //
+
+    // Returns the number of tokens in the KV cache
+    LLAMA_API DEPRECATED(int llama_get_kv_cache_token_count(const struct llama_context * ctx),
+            "avoid using this, it will be removed in the future, instead - count the tokens in user code");
+
+    // Remove all tokens data of cells in [c0, c1)
+    // c0 < 0 : [0,  c1]
+    // c1 < 0 : [c0, inf)
+    LLAMA_API void llama_kv_cache_tokens_rm(
+            struct llama_context * ctx,
+                         int32_t   c0,
+                         int32_t   c1);
+
+    // Removes all tokens that belong to the specified sequence and have positions in [p0, p1)
+    // p0 < 0 : [0,  p1]
+    // p1 < 0 : [p0, inf)
+    LLAMA_API void llama_kv_cache_seq_rm(
+            struct llama_context * ctx,
+                    llama_seq_id   seq_id,
+                       llama_pos   p0,
+                       llama_pos   p1);
+
+    // Copy all tokens that belong to the specified sequence to another sequence
+    // Note that this does not allocate extra KV cache memory - it simply assigns the tokens to the new sequence
+    // p0 < 0 : [0,  p1]
+    // p1 < 0 : [p0, inf)
+    LLAMA_API void llama_kv_cache_seq_cp(
+            struct llama_context * ctx,
+                    llama_seq_id   seq_id_src,
+                    llama_seq_id   seq_id_dst,
+                       llama_pos   p0,
+                       llama_pos   p1);
+
+    // Removes all tokens that do not belong to the specified sequence
+    LLAMA_API void llama_kv_cache_seq_keep(
+            struct llama_context * ctx,
+                    llama_seq_id   seq_id);
+
+    // Adds relative position "delta" to all tokens that belong to the specified sequence and have positions in [p0, p1)
+    // If the KV cache is RoPEd, the KV data is updated accordingly
+    // p0 < 0 : [0,  p1]
+    // p1 < 0 : [p0, inf)
+    LLAMA_API void llama_kv_cache_seq_shift(
+            struct llama_context * ctx,
+                    llama_seq_id   seq_id,
+                       llama_pos   p0,
+                       llama_pos   p1,
+                       llama_pos   delta);
+
+    //
+    // State / sessions
+    //
+
+    // Returns the maximum size in bytes of the state (rng, logits, embedding
+    // and kv_cache) - will often be smaller after compacting tokens
+    LLAMA_API size_t llama_get_state_size(const struct llama_context * ctx);
+
+    // Copies the state to the specified destination address.
+    // Destination needs to have allocated enough memory.
+    // Returns the number of bytes copied
+    LLAMA_API size_t llama_copy_state_data(
+            struct llama_context * ctx,
+                         uint8_t * dst);
+
+    // Set the state reading from the specified address
+    // Returns the number of bytes read
+    LLAMA_API size_t llama_set_state_data(
+            struct llama_context * ctx,
+                         uint8_t * src);
+
+    // Save/load session file
+    LLAMA_API bool llama_load_session_file(
+            struct llama_context * ctx,
+                      const char * path_session,
+                     llama_token * tokens_out,
+                          size_t   n_token_capacity,
+                          size_t * n_token_count_out);
+
+    LLAMA_API bool llama_save_session_file(
+            struct llama_context * ctx,
+                      const char * path_session,
+               const llama_token * tokens,
+                          size_t   n_token_count);
+
+    //
+    // Decoding
+    //
+
+    // Run the llama inference to obtain the logits and probabilities for the next token(s).
+    // tokens + n_tokens is the provided batch of new tokens to process
+    // n_past is the number of tokens to use from previous eval calls
+    // Returns 0 on success
+    // DEPRECATED: use llama_decode() instead
+    LLAMA_API DEPRECATED(int llama_eval(
+            struct llama_context * ctx,
+                     llama_token * tokens,
+                         int32_t   n_tokens,
+                             int   n_past),
+            "use llama_decode() instead");
+
+    // Same as llama_eval, but use float matrix input directly.
+    // DEPRECATED: use llama_decode() instead
+    LLAMA_API DEPRECATED(int llama_eval_embd(
+            struct llama_context * ctx,
+                           float * embd,
+                         int32_t   n_tokens,
+                             int   n_past),
+            "use llama_decode() instead");
+
+    // Return batch for single sequence of tokens starting at pos_0
+    //
+    // NOTE: this is a helper function to facilitate transition to the new batch API - avoid using it
+    //
+    LLAMA_API struct llama_batch llama_batch_get_one(
+                  llama_token * tokens,
+                      int32_t   n_tokens,
+                    llama_pos   pos_0,
+                 llama_seq_id   seq_id);
+
+    // Allocates a batch of tokens on the heap that can hold a maximum of n_tokens
+    // Each token can be assigned up to n_seq_max sequence ids
+    // The batch has to be freed with llama_batch_free()
+    // If embd != 0, llama_batch.embd will be allocated with size of n_tokens * embd * sizeof(float)
+    // Otherwise, llama_batch.token will be allocated to store n_tokens llama_token
+    // The rest of the llama_batch members are allocated with size n_tokens
+    // All members are left uninitialized
+    LLAMA_API struct llama_batch llama_batch_init(
+            int32_t n_tokens,
+            int32_t embd,
+            int32_t n_seq_max);
+
+    // Frees a batch of tokens allocated with llama_batch_init()
+    LLAMA_API void llama_batch_free(struct llama_batch batch);
+
+    // Positive return values does not mean a fatal error, but rather a warning.
+    //   0 - success
+    //   1 - could not find a KV slot for the batch (try reducing the size of the batch or increase the context)
+    // < 0 - error
+    LLAMA_API int llama_decode(
+            struct llama_context * ctx,
+              struct llama_batch   batch);
+
+    // Set the number of threads used for decoding
+    // n_threads is the number of threads used for generation (single token)
+    // n_threads_batch is the number of threads used for prompt and batch processing (multiple tokens)
+    LLAMA_API void llama_set_n_threads(struct llama_context * ctx, uint32_t n_threads, uint32_t n_threads_batch);
+
+    // Token logits obtained from the last call to llama_eval()
+    // The logits for the last token are stored in the last row
+    // Logits for which llama_batch.logits[i] == 0 are undefined
+    // Rows: n_tokens provided with llama_batch
+    // Cols: n_vocab
+    LLAMA_API float * llama_get_logits(struct llama_context * ctx);
+
+    // Logits for the ith token. Equivalent to:
+    // llama_get_logits(ctx) + i*n_vocab
+    LLAMA_API float * llama_get_logits_ith(struct llama_context * ctx, int32_t i);
+
+    // Get the embeddings for the input
+    // shape: [n_embd] (1-dimensional)
+    LLAMA_API float * llama_get_embeddings(struct llama_context * ctx);
+
+    //
+    // Vocab
+    //
+
+    LLAMA_API const char * llama_token_get_text(const struct llama_context * ctx, llama_token token);
+
+    LLAMA_API float llama_token_get_score(const struct llama_context * ctx, llama_token token);
+
+    LLAMA_API enum llama_token_type llama_token_get_type(const struct llama_context * ctx, llama_token token);
+
+    // Special tokens
+    LLAMA_API llama_token llama_token_bos(const struct llama_context * ctx);  // beginning-of-sentence
+    LLAMA_API llama_token llama_token_eos(const struct llama_context * ctx);  // end-of-sentence
+    LLAMA_API llama_token llama_token_nl (const struct llama_context * ctx);  // next-line
+    // codellama infill tokens
+    LLAMA_API llama_token llama_token_prefix(const struct llama_context * ctx); // Beginning of infill prefix
+    LLAMA_API llama_token llama_token_middle(const struct llama_context * ctx); // Beginning of infill middle
+    LLAMA_API llama_token llama_token_suffix(const struct llama_context * ctx); // Beginning of infill suffix
+    LLAMA_API llama_token llama_token_eot   (const struct llama_context * ctx); // End of infill middle
+
+    //
+    // Tokenization
+    //
+
+    /// @details Convert the provided text into tokens.
+    /// @param tokens The tokens pointer must be large enough to hold the resulting tokens.
+    /// @return Returns the number of tokens on success, no more than n_max_tokens
+    /// @return Returns a negative number on failure - the number of tokens that would have been returned
+    /// @param special Allow tokenizing special and/or control tokens which otherwise are not exposed and treated as plaintext.
+    ///                Does not insert a leading space.
+    LLAMA_API int llama_tokenize(
+        const struct llama_model * model,
+                      const char * text,
+                             int   text_len,
+                     llama_token * tokens,
+                             int   n_max_tokens,
+                            bool   add_bos,
+                            bool   special);
+
+    // Token Id -> Piece.
+    // Uses the vocabulary in the provided context.
+    // Does not write null terminator to the buffer.
+    // User code is responsible to remove the leading whitespace of the first non-BOS token when decoding multiple tokens.
+    LLAMA_API int llama_token_to_piece(
+              const struct llama_model * model,
+                           llama_token   token,
+                                  char * buf,
+                                  int    length);
+
+    //
+    // Grammar
+    //
+
+    LLAMA_API struct llama_grammar * llama_grammar_init(
+            const llama_grammar_element ** rules,
+                                 size_t    n_rules,
+                                 size_t    start_rule_index);
+
+    LLAMA_API void llama_grammar_free(struct llama_grammar * grammar);
+
+    LLAMA_API struct llama_grammar * llama_grammar_copy(const struct llama_grammar * grammar);
+
+    //
+    // Sampling functions
+    //
+
+    // Sets the current rng seed.
+    LLAMA_API void llama_set_rng_seed(struct llama_context * ctx, uint32_t seed);
+
+    /// @details Repetition penalty described in CTRL academic paper https://arxiv.org/abs/1909.05858, with negative logit fix.
+    /// @details Frequency and presence penalties described in OpenAI API https://platform.openai.com/docs/api-reference/parameter-details.
+    LLAMA_API void llama_sample_repetition_penalties(
+            struct llama_context * ctx,
+          llama_token_data_array * candidates,
+               const llama_token * last_tokens,
+                          size_t   penalty_last_n,
+                           float   penalty_repeat,
+                           float   penalty_freq,
+                           float   penalty_present);
+
+    /// @details Apply classifier-free guidance to the logits as described in academic paper "Stay on topic with Classifier-Free Guidance" https://arxiv.org/abs/2306.17806
+    /// @param candidates A vector of `llama_token_data` containing the candidate tokens, the logits must be directly extracted from the original generation context without being sorted.
+    /// @params guidance_ctx A separate context from the same model. Other than a negative prompt at the beginning, it should have all generated and user input tokens copied from the main context.
+    /// @params scale Guidance strength. 1.0f means no guidance. Higher values mean stronger guidance.
+    LLAMA_API void llama_sample_classifier_free_guidance(
+              struct llama_context * ctx,
+            llama_token_data_array * candidates,
+              struct llama_context * guidance_ctx,
+                             float   scale);
+
+    /// @details Sorts candidate tokens by their logits in descending order and calculate probabilities based on logits.
+    LLAMA_API void llama_sample_softmax(
+            struct llama_context * ctx,
+          llama_token_data_array * candidates);
+
+    /// @details Top-K sampling described in academic paper "The Curious Case of Neural Text Degeneration" https://arxiv.org/abs/1904.09751
+    LLAMA_API void llama_sample_top_k(
+            struct llama_context * ctx,
+          llama_token_data_array * candidates,
+                             int   k,
+                          size_t   min_keep);
+
+    /// @details Nucleus sampling described in academic paper "The Curious Case of Neural Text Degeneration" https://arxiv.org/abs/1904.09751
+    LLAMA_API void llama_sample_top_p(
+            struct llama_context * ctx,
+          llama_token_data_array * candidates,
+                           float   p,
+                          size_t   min_keep);
+
+    /// @details Tail Free Sampling described in https://www.trentonbricken.com/Tail-Free-Sampling/.
+    LLAMA_API void llama_sample_tail_free(
+            struct llama_context * ctx,
+          llama_token_data_array * candidates,
+                           float   z,
+                          size_t   min_keep);
+
+    /// @details Locally Typical Sampling implementation described in the paper https://arxiv.org/abs/2202.00666.
+    LLAMA_API void llama_sample_typical(
+            struct llama_context * ctx,
+          llama_token_data_array * candidates,
+                           float   p,
+                          size_t   min_keep);
+
+    LLAMA_API void llama_sample_temp(
+            struct llama_context * ctx,
+          llama_token_data_array * candidates,
+                           float   temp);
+
+    LLAMA_API DEPRECATED(void llama_sample_temperature(
+                struct llama_context * ctx,
+              llama_token_data_array * candidates,
+                               float   temp),
+            "use llama_sample_temp instead");
+
+    /// @details Apply constraints from grammar
+    LLAMA_API void llama_sample_grammar(
+            struct llama_context * ctx,
+          llama_token_data_array * candidates,
+      const struct llama_grammar * grammar);
+
+    /// @details Mirostat 1.0 algorithm described in the paper https://arxiv.org/abs/2007.14966. Uses tokens instead of words.
+    /// @param candidates A vector of `llama_token_data` containing the candidate tokens, their probabilities (p), and log-odds (logit) for the current position in the generated text.
+    /// @param tau  The target cross-entropy (or surprise) value you want to achieve for the generated text. A higher value corresponds to more surprising or less predictable text, while a lower value corresponds to less surprising or more predictable text.
+    /// @param eta The learning rate used to update `mu` based on the error between the target and observed surprisal of the sampled word. A larger learning rate will cause `mu` to be updated more quickly, while a smaller learning rate will result in slower updates.
+    /// @param m The number of tokens considered in the estimation of `s_hat`. This is an arbitrary value that is used to calculate `s_hat`, which in turn helps to calculate the value of `k`. In the paper, they use `m = 100`, but you can experiment with different values to see how it affects the performance of the algorithm.
+    /// @param mu Maximum cross-entropy. This value is initialized to be twice the target cross-entropy (`2 * tau`) and is updated in the algorithm based on the error between the target and observed surprisal.
+    LLAMA_API llama_token llama_sample_token_mirostat(
+            struct llama_context * ctx,
+          llama_token_data_array * candidates,
+                           float   tau,
+                           float   eta,
+                             int   m,
+                           float * mu);
+
+    /// @details Mirostat 2.0 algorithm described in the paper https://arxiv.org/abs/2007.14966. Uses tokens instead of words.
+    /// @param candidates A vector of `llama_token_data` containing the candidate tokens, their probabilities (p), and log-odds (logit) for the current position in the generated text.
+    /// @param tau  The target cross-entropy (or surprise) value you want to achieve for the generated text. A higher value corresponds to more surprising or less predictable text, while a lower value corresponds to less surprising or more predictable text.
+    /// @param eta The learning rate used to update `mu` based on the error between the target and observed surprisal of the sampled word. A larger learning rate will cause `mu` to be updated more quickly, while a smaller learning rate will result in slower updates.
+    /// @param mu Maximum cross-entropy. This value is initialized to be twice the target cross-entropy (`2 * tau`) and is updated in the algorithm based on the error between the target and observed surprisal.
+    LLAMA_API llama_token llama_sample_token_mirostat_v2(
+            struct llama_context * ctx,
+          llama_token_data_array * candidates,
+                           float   tau,
+                           float   eta,
+                           float * mu);
+
+    /// @details Selects the token with the highest probability.
+    LLAMA_API llama_token llama_sample_token_greedy(
+            struct llama_context * ctx,
+          llama_token_data_array * candidates);
+
+    /// @details Randomly selects a token from the candidates based on their probabilities.
+    LLAMA_API llama_token llama_sample_token(
+            struct llama_context * ctx,
+          llama_token_data_array * candidates);
+
+    /// @details Accepts the sampled token into the grammar
+    LLAMA_API void llama_grammar_accept_token(
+            struct llama_context * ctx,
+            struct llama_grammar * grammar,
+                     llama_token   token);
+
+    //
+    // Beam search
+    //
+
+    struct llama_beam_view {
+        const llama_token * tokens;
+
+        size_t n_tokens;
+        float  p;        // Cumulative beam probability (renormalized relative to all beams)
+        bool   eob;      // Callback should set this to true when a beam is at end-of-beam.
+    };
+
+    // Passed to beam_search_callback function.
+    // Whenever 0 < common_prefix_length, this number of tokens should be copied from any of the beams
+    // (e.g. beams[0]) as they will be removed (shifted) from all beams in all subsequent callbacks.
+    // These pointers are valid only during the synchronous callback, so should not be saved.
+    struct llama_beams_state {
+        struct llama_beam_view * beam_views;
+
+        size_t n_beams;               // Number of elements in beam_views[].
+        size_t common_prefix_length;  // Current max length of prefix tokens shared by all beams.
+        bool   last_call;             // True iff this is the last callback invocation.
+    };
+
+    // Type of pointer to the beam_search_callback function.
+    // void* callback_data is any custom data passed to llama_beam_search, that is subsequently
+    // passed back to beam_search_callback. This avoids having to use global variables in the callback.
+    typedef void (*llama_beam_search_callback_fn_t)(void * callback_data, struct llama_beams_state);
+
+    /// @details Deterministically returns entire sentence constructed by a beam search.
+    /// @param ctx Pointer to the llama_context.
+    /// @param callback Invoked for each iteration of the beam_search loop, passing in beams_state.
+    /// @param callback_data A pointer that is simply passed back to callback.
+    /// @param n_beams Number of beams to use.
+    /// @param n_past Number of tokens already evaluated.
+    /// @param n_predict Maximum number of tokens to predict. EOS may occur earlier.
+    LLAMA_API void llama_beam_search(
+                   struct llama_context * ctx,
+        llama_beam_search_callback_fn_t   callback,
+                                   void * callback_data,
+                                 size_t   n_beams,
+                                    int   n_past,
+                                    int   n_predict);
+
+    // Performance information
+    LLAMA_API struct llama_timings llama_get_timings(struct llama_context * ctx);
+
+    LLAMA_API void llama_print_timings(struct llama_context * ctx);
+    LLAMA_API void llama_reset_timings(struct llama_context * ctx);
+
+    // Print system information
+    LLAMA_API const char * llama_print_system_info(void);
+
+    // Set callback for all future logging events.
+    // If this is not called, or NULL is supplied, everything is output on stderr.
+    LLAMA_API void llama_log_set(ggml_log_callback log_callback, void * user_data);
+
+    LLAMA_API void llama_dump_timing_info_yaml(FILE * stream, const struct llama_context * ctx);
+
+#ifdef __cplusplus
+}
+#endif
+
+// Internal API to be implemented by llama.cpp and used by tests/benchmarks only
+#ifdef LLAMA_API_INTERNAL
+
+#include <vector>
+#include <string>
+
+struct ggml_tensor;
+
+const std::vector<std::pair<std::string, struct ggml_tensor *>> & llama_internal_get_tensor_map(
+    struct llama_context * ctx
+);
+
+#endif // LLAMA_API_INTERNAL
+
+#endif // LLAMA_H

runner/main.cpp

@@ -0,0 +1,272 @@
+#include <cmath>
+#include <cstdio>
+#include <string>
+#include <vector>
+#include <thread>
+
+#include "llama.h"
+#include "main.h"
+
+std::vector<llama_token> tokenize(const struct llama_model * model, const std::string & text, bool add_bos, bool special = false) {
+    // upper limit for the number of tokens
+    int n_tokens = text.length() + add_bos;
+    std::vector<llama_token> result(n_tokens);
+    n_tokens = llama_tokenize(model, text.data(), text.length(), result.data(), result.size(), add_bos, special);
+    if (n_tokens < 0) {
+        result.resize(-n_tokens);
+        int check = llama_tokenize(model, text.data(), text.length(), result.data(), result.size(), add_bos, special);
+        GGML_ASSERT(check == -n_tokens);
+    } else {
+        result.resize(n_tokens);
+    }
+    return result;
+}
+
+std::string token_to_piece(const struct llama_context * ctx, llama_token token) {
+    std::vector<char> result(8, 0);
+    const int n_tokens = llama_token_to_piece(llama_get_model(ctx), token, result.data(), result.size());
+    if (n_tokens < 0) {
+        result.resize(-n_tokens);
+        int check = llama_token_to_piece(llama_get_model(ctx), token, result.data(), result.size());
+        GGML_ASSERT(check == -n_tokens);
+    } else {
+        result.resize(n_tokens);
+    }
+
+    return std::string(result.data(), result.size());
+}
+
+void batch_add(
+                 struct llama_batch & batch,
+                        llama_token   id,
+                          llama_pos   pos,
+    const std::vector<llama_seq_id> & seq_ids,
+                               bool   logits) {
+    batch.token   [batch.n_tokens] = id;
+    batch.pos     [batch.n_tokens] = pos,
+    batch.n_seq_id[batch.n_tokens] = seq_ids.size();
+    for (size_t i = 0; i < seq_ids.size(); ++i) {
+        batch.seq_id[batch.n_tokens][i] = seq_ids[i];
+    }
+    batch.logits  [batch.n_tokens] = logits;
+
+    batch.n_tokens++;
+}
+
+void batch_clear(struct llama_batch & batch) {
+    batch.n_tokens = 0;
+}
+
+int generate(const char *model_path, const char *prompt) {
+    // number of parallel batches
+    int n_parallel = 1;
+
+    // total length of the sequences including the prompt
+    int n_len = 32;
+
+    // init LLM
+    llama_backend_init(true);
+
+    // initialize the model
+
+    llama_model_params model_params = llama_model_default_params();
+
+    // model_params.n_gpu_layers = 99; // offload all layers to the GPU
+
+    llama_model * model = llama_load_model_from_file(model_path, model_params);
+
+    if (model == NULL) {
+        fprintf(stderr , "%s: error: unable to load model\n" , __func__);
+        return 1;
+    }
+
+    // tokenize the prompt
+
+    std::vector<llama_token> tokens_list;
+    tokens_list = tokenize(model, std::string(prompt), true);
+    const int n_kv_req = tokens_list.size() + (n_len - tokens_list.size())*n_parallel;
+
+    // initialize the context
+
+    llama_context_params ctx_params = llama_context_default_params();
+
+    ctx_params.seed  = 1234;
+    ctx_params.n_ctx = n_kv_req;
+    ctx_params.n_batch = std::max(n_len, n_parallel);
+    ctx_params.n_threads       = std::thread::hardware_concurrency();
+    ctx_params.n_threads_batch = ctx_params.n_threads;
+
+    llama_context * ctx = llama_new_context_with_model(model, ctx_params);
+
+    if (ctx == NULL) {
+        fprintf(stderr , "%s: error: failed to create the llama_context\n" , __func__);
+        return 1;
+    }
+
+    const int n_ctx    = llama_n_ctx(ctx);
+
+    printf("\n%s: n_len = %d, n_ctx = %d, n_batch = %d, n_parallel = %d, n_kv_req = %d\n", __func__, n_len, n_ctx, ctx_params.n_batch, n_parallel, n_kv_req);
+
+    // make sure the KV cache is big enough to hold all the prompt and generated tokens
+    if (n_kv_req > n_ctx) {
+        printf("%s: error: n_kv_req (%d) > n_ctx, the required KV cache size is not big enough\n", __func__,  n_kv_req);
+        printf("%s:        either reduce n_parallel or increase n_ctx\n", __func__);
+        return 1;
+    }
+
+    // print the prompt token-by-token
+
+    fprintf(stderr, "\n");
+
+    for (auto id : tokens_list) {
+        fprintf(stderr, "%s", token_to_piece(ctx, id).c_str());
+    }
+
+    fflush(stderr);
+
+    // create a llama_batch
+    // we use this object to submit token data for decoding
+    llama_batch batch = llama_batch_init(std::max(tokens_list.size(), (size_t)n_parallel), 0, 1);
+
+    // evaluate the initial prompt
+    for (size_t i = 0; i < tokens_list.size(); ++i) {
+        batch_add(batch, tokens_list[i], i, { 0 }, false);
+    }
+    GGML_ASSERT(batch.n_tokens == (int) tokens_list.size());
+
+    // llama_decode will output logits only for the last token of the prompt
+    batch.logits[batch.n_tokens - 1] = true;
+
+    if (llama_decode(ctx, batch) != 0) {
+        printf("%s: llama_decode() failed\n", __func__);
+        return 1;
+    }
+
+    // assign the system KV cache to all parallel sequences
+    // this way, the parallel sequences will "reuse" the prompt tokens without having to copy them
+    for (int32_t i = 1; i < n_parallel; ++i) {
+        llama_kv_cache_seq_cp(ctx, 0, i, 0, batch.n_tokens);
+    }
+
+    if (n_parallel > 1) {
+        printf("\n\n%s: generating %d sequences ...\n", __func__, n_parallel);
+    }
+
+    // main loop
+
+    // we will store the parallel decoded sequences in this vector
+    std::vector<std::string> streams(n_parallel);
+
+    // remember the batch index of the last token for each parallel sequence
+    // we need this to determine which logits to sample from
+    std::vector<int32_t> i_batch(n_parallel, batch.n_tokens - 1);
+
+    int n_cur    = batch.n_tokens;
+    int n_decode = 0;
+
+    const auto t_main_start = ggml_time_us();
+
+    while (n_cur <= n_len) {
+        // prepare the next batch
+        batch_clear(batch);
+
+        // sample the next token for each parallel sequence / stream
+        for (int32_t i = 0; i < n_parallel; ++i) {
+            if (i_batch[i] < 0) {
+                // the stream has already finished
+                continue;
+            }
+
+            auto   n_vocab = llama_n_vocab(model);
+            auto * logits  = llama_get_logits_ith(ctx, i_batch[i]);
+
+            std::vector<llama_token_data> candidates;
+            candidates.reserve(n_vocab);
+
+            for (llama_token token_id = 0; token_id < n_vocab; token_id++) {
+                candidates.emplace_back(llama_token_data{ token_id, logits[token_id], 0.0f });
+            }
+
+            llama_token_data_array candidates_p = { candidates.data(), candidates.size(), false };
+
+            const int   top_k = 40;
+            const float top_p = 0.9f;
+            const float temp  = 0.4f;
+
+            llama_sample_top_k(ctx, &candidates_p, top_k, 1);
+            llama_sample_top_p(ctx, &candidates_p, top_p, 1);
+            llama_sample_temp (ctx, &candidates_p, temp);
+
+            const llama_token new_token_id = llama_sample_token(ctx, &candidates_p);
+
+            //const llama_token new_token_id = llama_sample_token_greedy(ctx, &candidates_p);
+
+            // is it an end of stream? -> mark the stream as finished
+            if (new_token_id == llama_token_eos(ctx) || n_cur == n_len) {
+                i_batch[i] = -1;
+                printf("\n");
+                if (n_parallel > 1) {
+                    printf("%s: stream %d finished at n_cur = %d", __func__, i, n_cur);
+                }
+
+                continue;
+            }
+
+            // if there is only one stream, we print immediately to stdout
+            if (n_parallel == 1) {
+                printf("%s", token_to_piece(ctx, new_token_id).c_str());
+                fflush(stdout);
+            }
+
+            streams[i] += token_to_piece(ctx, new_token_id);
+
+            i_batch[i] = batch.n_tokens;
+
+            // push this new token for next evaluation
+            batch_add(batch, new_token_id, n_cur, { i }, true);
+
+            n_decode += 1;
+        }
+
+        // all streams are finished
+        if (batch.n_tokens == 0) {
+            break;
+        }
+
+        n_cur += 1;
+
+        // evaluate the current batch with the transformer model
+        if (llama_decode(ctx, batch)) {
+            fprintf(stderr, "%s : failed to eval, return code %d\n", __func__, 1);
+            return 1;
+        }
+    }
+
+    printf("\n");
+
+    if (n_parallel > 1) {
+        printf("\n");
+
+        for (int32_t i = 0; i < n_parallel; ++i) {
+            printf("sequence %d:\n\n%s%s\n\n", i, prompt, streams[i].c_str());
+        }
+    }
+
+    const auto t_main_end = ggml_time_us();
+
+    printf("%s: decoded %d tokens in %.2f s, speed: %.2f t/s\n",
+            __func__, n_decode, (t_main_end - t_main_start) / 1000000.0f, n_decode / ((t_main_end - t_main_start) / 1000000.0f));
+
+    llama_print_timings(ctx);
+
+    fprintf(stderr, "\n");
+
+    llama_batch_free(batch);
+
+    llama_free(ctx);
+    llama_free_model(model);
+
+    llama_backend_free();
+
+    return 0;
+}

runner/main.go

@@ -0,0 +1,32 @@
+package main
+
+/*
+#cgo CFLAGS: -Ofast -std=c11 -fPIC -Wno-deprecated-declarations -Wno-unused-but-set-variable
+#cgo CPPFLAGS: -Ofast -Wall -Wextra -Wno-unused-function -Wno-unused-variable -Wno-deprecated-declarations -Wno-unused-but-set-variable -DNDEBUG -DGGML_USE_K_QUANTS
+#cgo CXXFLAGS: -std=c++11 -fPIC
+#cgo darwin CPPFLAGS:  -DGGML_USE_ACCELERATE
+#cgo darwin,arm64 CPPFLAGS: -DGGML_USE_METAL
+#cgo darwin LDFLAGS: -framework Accelerate -framework Foundation -framework Metal -framework MetalKit -framework MetalPerformanceShaders
+#include <stdlib.h>
+#include "main.h"
+*/
+import "C"
+import (
+	"fmt"
+	"os"
+	"unsafe"
+)
+
+func main() {
+	if len(os.Args) < 2 {
+		fmt.Println("No prompt provided")
+		return
+	}
+
+	prompt := C.CString(os.Args[1])
+	defer C.free(unsafe.Pointer(prompt))
+	model := C.CString("./model.bin")
+	defer C.free(unsafe.Pointer(model))
+
+	C.generate(model, prompt)
+}

runner/main.h

@@ -0,0 +1,14 @@
+#ifndef MAIN_H
+#define MAIN_H
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+int generate(const char *model_path, const char *prompt);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif // MAIN_H

runner/unicode.h

@@ -0,0 +1,488 @@
+/**
+ * llama.cpp - git 465219b9143ac01db0990bbcb0a081ef72ec2008
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023 Georgi Gerganov
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#pragma once
+
+#include <cassert>
+#include <stdexcept>
+#include <vector>
+#include <unordered_map>
+
+static const std::vector<std::pair<uint32_t, uint32_t>> digit_ranges = {
+{0x30, 0x39}, {0xB2, 0xB3}, {0xB9, 0xB9}, {0x660, 0x669}, {0x6F0, 0x6F9}, {0x7C0, 0x7C9}, {0x966, 0x96F}, {0x9E6, 0x9EF}, {0xA66, 0xA6F}, {0xAE6, 0xAEF}, {0xB66, 0xB6F}, {0xBE6, 0xBEF}, {0xC66, 0xC6F},
+{0xCE6, 0xCEF}, {0xD66, 0xD6F}, {0xDE6, 0xDEF}, {0xE50, 0xE59}, {0xED0, 0xED9}, {0xF20, 0xF29}, {0x1040, 0x1049}, {0x1090, 0x1099}, {0x1369, 0x1371}, {0x17E0, 0x17E9}, {0x1810, 0x1819}, {0x1946, 0x194F},
+{0x19D0, 0x19DA}, {0x1A80, 0x1A89}, {0x1A90, 0x1A99}, {0x1B50, 0x1B59}, {0x1BB0, 0x1BB9}, {0x1C40, 0x1C49}, {0x1C50, 0x1C59}, {0x2070, 0x2070}, {0x2074, 0x2079}, {0x2080, 0x2089}, {0x2460, 0x2468},
+{0x2474, 0x247C}, {0x2488, 0x2490}, {0x24EA, 0x24EA}, {0x24F5, 0x24FD}, {0x24FF, 0x24FF}, {0x2776, 0x277E}, {0x2780, 0x2788}, {0x278A, 0x2792}, {0xA620, 0xA629}, {0xA8D0, 0xA8D9}, {0xA900, 0xA909},
+{0xA9D0, 0xA9D9}, {0xA9F0, 0xA9F9}, {0xAA50, 0xAA59}, {0xABF0, 0xABF9}, {0xFF10, 0xFF19}, {0x104A0, 0x104A9}, {0x10A40, 0x10A43}, {0x10D30, 0x10D39}, {0x10E60, 0x10E68}, {0x11052, 0x1105A},
+{0x11066, 0x1106F}, {0x110F0, 0x110F9}, {0x11136, 0x1113F}, {0x111D0, 0x111D9}, {0x112F0, 0x112F9}, {0x11450, 0x11459}, {0x114D0, 0x114D9}, {0x11650, 0x11659}, {0x116C0, 0x116C9}, {0x11730, 0x11739},
+{0x118E0, 0x118E9}, {0x11950, 0x11959}, {0x11C50, 0x11C59}, {0x11D50, 0x11D59}, {0x11DA0, 0x11DA9}, {0x16A60, 0x16A69}, {0x16B50, 0x16B59}, {0x1D7CE, 0x1D7FF}, {0x1E140, 0x1E149}, {0x1E2F0, 0x1E2F9},
+{0x1E950, 0x1E959}, {0x1F100, 0x1F10A}, {0x1FBF0, 0x1FBF9},
+};
+
+static const std::vector<std::pair<uint32_t, uint32_t>> letter_ranges = {
+{0x41, 0x5A}, {0x61, 0x7A}, {0xAA, 0xAA}, {0xB5, 0xB5}, {0xBA, 0xBA}, {0xC0, 0xD6}, {0xD8, 0xF6}, {0xF8, 0x2C1}, {0x2C6, 0x2D1}, {0x2E0, 0x2E4}, {0x2EC, 0x2EC}, {0x2EE, 0x2EE}, {0x370, 0x374},
+{0x376, 0x377}, {0x37A, 0x37D}, {0x37F, 0x37F}, {0x386, 0x386}, {0x388, 0x38A}, {0x38C, 0x38C}, {0x38E, 0x3A1}, {0x3A3, 0x3F5}, {0x3F7, 0x481}, {0x48A, 0x52F}, {0x531, 0x556}, {0x559, 0x559},
+{0x560, 0x588}, {0x5D0, 0x5EA}, {0x5EF, 0x5F2}, {0x620, 0x64A}, {0x66E, 0x66F}, {0x671, 0x6D3}, {0x6D5, 0x6D5}, {0x6E5, 0x6E6}, {0x6EE, 0x6EF}, {0x6FA, 0x6FC}, {0x6FF, 0x6FF}, {0x710, 0x710},
+{0x712, 0x72F}, {0x74D, 0x7A5}, {0x7B1, 0x7B1}, {0x7CA, 0x7EA}, {0x7F4, 0x7F5}, {0x7FA, 0x7FA}, {0x800, 0x815}, {0x81A, 0x81A}, {0x824, 0x824}, {0x828, 0x828}, {0x840, 0x858}, {0x860, 0x86A},
+{0x8A0, 0x8B4}, {0x8B6, 0x8C7}, {0x904, 0x939}, {0x93D, 0x93D}, {0x950, 0x950}, {0x958, 0x961}, {0x971, 0x980}, {0x985, 0x98C}, {0x98F, 0x990}, {0x993, 0x9A8}, {0x9AA, 0x9B0}, {0x9B2, 0x9B2},
+{0x9B6, 0x9B9}, {0x9BD, 0x9BD}, {0x9CE, 0x9CE}, {0x9DC, 0x9DD}, {0x9DF, 0x9E1}, {0x9F0, 0x9F1}, {0x9FC, 0x9FC}, {0xA05, 0xA0A}, {0xA0F, 0xA10}, {0xA13, 0xA28}, {0xA2A, 0xA30}, {0xA32, 0xA33},
+{0xA35, 0xA36}, {0xA38, 0xA39}, {0xA59, 0xA5C}, {0xA5E, 0xA5E}, {0xA72, 0xA74}, {0xA85, 0xA8D}, {0xA8F, 0xA91}, {0xA93, 0xAA8}, {0xAAA, 0xAB0}, {0xAB2, 0xAB3}, {0xAB5, 0xAB9}, {0xABD, 0xABD},
+{0xAD0, 0xAD0}, {0xAE0, 0xAE1}, {0xAF9, 0xAF9}, {0xB05, 0xB0C}, {0xB0F, 0xB10}, {0xB13, 0xB28}, {0xB2A, 0xB30}, {0xB32, 0xB33}, {0xB35, 0xB39}, {0xB3D, 0xB3D}, {0xB5C, 0xB5D}, {0xB5F, 0xB61},
+{0xB71, 0xB71}, {0xB83, 0xB83}, {0xB85, 0xB8A}, {0xB8E, 0xB90}, {0xB92, 0xB95}, {0xB99, 0xB9A}, {0xB9C, 0xB9C}, {0xB9E, 0xB9F}, {0xBA3, 0xBA4}, {0xBA8, 0xBAA}, {0xBAE, 0xBB9}, {0xBD0, 0xBD0},
+{0xC05, 0xC0C}, {0xC0E, 0xC10}, {0xC12, 0xC28}, {0xC2A, 0xC39}, {0xC3D, 0xC3D}, {0xC58, 0xC5A}, {0xC60, 0xC61}, {0xC80, 0xC80}, {0xC85, 0xC8C}, {0xC8E, 0xC90}, {0xC92, 0xCA8}, {0xCAA, 0xCB3},
+{0xCB5, 0xCB9}, {0xCBD, 0xCBD}, {0xCDE, 0xCDE}, {0xCE0, 0xCE1}, {0xCF1, 0xCF2}, {0xD04, 0xD0C}, {0xD0E, 0xD10}, {0xD12, 0xD3A}, {0xD3D, 0xD3D}, {0xD4E, 0xD4E}, {0xD54, 0xD56}, {0xD5F, 0xD61},
+{0xD7A, 0xD7F}, {0xD85, 0xD96}, {0xD9A, 0xDB1}, {0xDB3, 0xDBB}, {0xDBD, 0xDBD}, {0xDC0, 0xDC6}, {0xE01, 0xE30}, {0xE32, 0xE33}, {0xE40, 0xE46}, {0xE81, 0xE82}, {0xE84, 0xE84}, {0xE86, 0xE8A},
+{0xE8C, 0xEA3}, {0xEA5, 0xEA5}, {0xEA7, 0xEB0}, {0xEB2, 0xEB3}, {0xEBD, 0xEBD}, {0xEC0, 0xEC4}, {0xEC6, 0xEC6}, {0xEDC, 0xEDF}, {0xF00, 0xF00}, {0xF40, 0xF47}, {0xF49, 0xF6C}, {0xF88, 0xF8C},
+{0x1000, 0x102A}, {0x103F, 0x103F}, {0x1050, 0x1055}, {0x105A, 0x105D}, {0x1061, 0x1061}, {0x1065, 0x1066}, {0x106E, 0x1070}, {0x1075, 0x1081}, {0x108E, 0x108E}, {0x10A0, 0x10C5}, {0x10C7, 0x10C7},
+{0x10CD, 0x10CD}, {0x10D0, 0x10FA}, {0x10FC, 0x1248}, {0x124A, 0x124D}, {0x1250, 0x1256}, {0x1258, 0x1258}, {0x125A, 0x125D}, {0x1260, 0x1288}, {0x128A, 0x128D}, {0x1290, 0x12B0}, {0x12B2, 0x12B5},
+{0x12B8, 0x12BE}, {0x12C0, 0x12C0}, {0x12C2, 0x12C5}, {0x12C8, 0x12D6}, {0x12D8, 0x1310}, {0x1312, 0x1315}, {0x1318, 0x135A}, {0x1380, 0x138F}, {0x13A0, 0x13F5}, {0x13F8, 0x13FD}, {0x1401, 0x166C},
+{0x166F, 0x167F}, {0x1681, 0x169A}, {0x16A0, 0x16EA}, {0x16F1, 0x16F8}, {0x1700, 0x170C}, {0x170E, 0x1711}, {0x1720, 0x1731}, {0x1740, 0x1751}, {0x1760, 0x176C}, {0x176E, 0x1770}, {0x1780, 0x17B3},
+{0x17D7, 0x17D7}, {0x17DC, 0x17DC}, {0x1820, 0x1878}, {0x1880, 0x1884}, {0x1887, 0x18A8}, {0x18AA, 0x18AA}, {0x18B0, 0x18F5}, {0x1900, 0x191E}, {0x1950, 0x196D}, {0x1970, 0x1974}, {0x1980, 0x19AB},
+{0x19B0, 0x19C9}, {0x1A00, 0x1A16}, {0x1A20, 0x1A54}, {0x1AA7, 0x1AA7}, {0x1B05, 0x1B33}, {0x1B45, 0x1B4B}, {0x1B83, 0x1BA0}, {0x1BAE, 0x1BAF}, {0x1BBA, 0x1BE5}, {0x1C00, 0x1C23}, {0x1C4D, 0x1C4F},
+{0x1C5A, 0x1C7D}, {0x1C80, 0x1C88}, {0x1C90, 0x1CBA}, {0x1CBD, 0x1CBF}, {0x1CE9, 0x1CEC}, {0x1CEE, 0x1CF3}, {0x1CF5, 0x1CF6}, {0x1CFA, 0x1CFA}, {0x1D00, 0x1DBF}, {0x1E00, 0x1F15}, {0x1F18, 0x1F1D},
+{0x1F20, 0x1F45}, {0x1F48, 0x1F4D}, {0x1F50, 0x1F57}, {0x1F59, 0x1F59}, {0x1F5B, 0x1F5B}, {0x1F5D, 0x1F5D}, {0x1F5F, 0x1F7D}, {0x1F80, 0x1FB4}, {0x1FB6, 0x1FBC}, {0x1FBE, 0x1FBE}, {0x1FC2, 0x1FC4},
+{0x1FC6, 0x1FCC}, {0x1FD0, 0x1FD3}, {0x1FD6, 0x1FDB}, {0x1FE0, 0x1FEC}, {0x1FF2, 0x1FF4}, {0x1FF6, 0x1FFC}, {0x2071, 0x2071}, {0x207F, 0x207F}, {0x2090, 0x209C}, {0x2102, 0x2102}, {0x2107, 0x2107},
+{0x210A, 0x2113}, {0x2115, 0x2115}, {0x2119, 0x211D}, {0x2124, 0x2124}, {0x2126, 0x2126}, {0x2128, 0x2128}, {0x212A, 0x212D}, {0x212F, 0x2139}, {0x213C, 0x213F}, {0x2145, 0x2149}, {0x214E, 0x214E},
+{0x2183, 0x2184}, {0x2C00, 0x2C2E}, {0x2C30, 0x2C5E}, {0x2C60, 0x2CE4}, {0x2CEB, 0x2CEE}, {0x2CF2, 0x2CF3}, {0x2D00, 0x2D25}, {0x2D27, 0x2D27}, {0x2D2D, 0x2D2D}, {0x2D30, 0x2D67}, {0x2D6F, 0x2D6F},
+{0x2D80, 0x2D96}, {0x2DA0, 0x2DA6}, {0x2DA8, 0x2DAE}, {0x2DB0, 0x2DB6}, {0x2DB8, 0x2DBE}, {0x2DC0, 0x2DC6}, {0x2DC8, 0x2DCE}, {0x2DD0, 0x2DD6}, {0x2DD8, 0x2DDE}, {0x2E2F, 0x2E2F}, {0x3005, 0x3006},
+{0x3031, 0x3035}, {0x303B, 0x303C}, {0x3041, 0x3096}, {0x309D, 0x309F}, {0x30A1, 0x30FA}, {0x30FC, 0x30FF}, {0x3105, 0x312F}, {0x3131, 0x318E}, {0x31A0, 0x31BF}, {0x31F0, 0x31FF}, {0x3400, 0x4DBF},
+{0x4E00, 0x9FFC}, {0xA000, 0xA48C}, {0xA4D0, 0xA4FD}, {0xA500, 0xA60C}, {0xA610, 0xA61F}, {0xA62A, 0xA62B}, {0xA640, 0xA66E}, {0xA67F, 0xA69D}, {0xA6A0, 0xA6E5}, {0xA717, 0xA71F}, {0xA722, 0xA788},
+{0xA78B, 0xA7BF}, {0xA7C2, 0xA7CA}, {0xA7F5, 0xA801}, {0xA803, 0xA805}, {0xA807, 0xA80A}, {0xA80C, 0xA822}, {0xA840, 0xA873}, {0xA882, 0xA8B3}, {0xA8F2, 0xA8F7}, {0xA8FB, 0xA8FB}, {0xA8FD, 0xA8FE},
+{0xA90A, 0xA925}, {0xA930, 0xA946}, {0xA960, 0xA97C}, {0xA984, 0xA9B2}, {0xA9CF, 0xA9CF}, {0xA9E0, 0xA9E4}, {0xA9E6, 0xA9EF}, {0xA9FA, 0xA9FE}, {0xAA00, 0xAA28}, {0xAA40, 0xAA42}, {0xAA44, 0xAA4B},
+{0xAA60, 0xAA76}, {0xAA7A, 0xAA7A}, {0xAA7E, 0xAAAF}, {0xAAB1, 0xAAB1}, {0xAAB5, 0xAAB6}, {0xAAB9, 0xAABD}, {0xAAC0, 0xAAC0}, {0xAAC2, 0xAAC2}, {0xAADB, 0xAADD}, {0xAAE0, 0xAAEA}, {0xAAF2, 0xAAF4},
+{0xAB01, 0xAB06}, {0xAB09, 0xAB0E}, {0xAB11, 0xAB16}, {0xAB20, 0xAB26}, {0xAB28, 0xAB2E}, {0xAB30, 0xAB5A}, {0xAB5C, 0xAB69}, {0xAB70, 0xABE2}, {0xAC00, 0xD7A3}, {0xD7B0, 0xD7C6}, {0xD7CB, 0xD7FB},
+{0xF900, 0xFA6D}, {0xFA70, 0xFAD9}, {0xFB00, 0xFB06}, {0xFB13, 0xFB17}, {0xFB1D, 0xFB1D}, {0xFB1F, 0xFB28}, {0xFB2A, 0xFB36}, {0xFB38, 0xFB3C}, {0xFB3E, 0xFB3E}, {0xFB40, 0xFB41}, {0xFB43, 0xFB44},
+{0xFB46, 0xFBB1}, {0xFBD3, 0xFD3D}, {0xFD50, 0xFD8F}, {0xFD92, 0xFDC7}, {0xFDF0, 0xFDFB}, {0xFE70, 0xFE74}, {0xFE76, 0xFEFC}, {0xFF21, 0xFF3A}, {0xFF41, 0xFF5A}, {0xFF66, 0xFFBE}, {0xFFC2, 0xFFC7},
+{0xFFCA, 0xFFCF}, {0xFFD2, 0xFFD7}, {0xFFDA, 0xFFDC}, {0x10000, 0x1000B}, {0x1000D, 0x10026}, {0x10028, 0x1003A}, {0x1003C, 0x1003D}, {0x1003F, 0x1004D}, {0x10050, 0x1005D}, {0x10080, 0x100FA},
+{0x10280, 0x1029C}, {0x102A0, 0x102D0}, {0x10300, 0x1031F}, {0x1032D, 0x10340}, {0x10342, 0x10349}, {0x10350, 0x10375}, {0x10380, 0x1039D}, {0x103A0, 0x103C3}, {0x103C8, 0x103CF}, {0x10400, 0x1049D},
+{0x104B0, 0x104D3}, {0x104D8, 0x104FB}, {0x10500, 0x10527}, {0x10530, 0x10563}, {0x10600, 0x10736}, {0x10740, 0x10755}, {0x10760, 0x10767}, {0x10800, 0x10805}, {0x10808, 0x10808}, {0x1080A, 0x10835},
+{0x10837, 0x10838}, {0x1083C, 0x1083C}, {0x1083F, 0x10855}, {0x10860, 0x10876}, {0x10880, 0x1089E}, {0x108E0, 0x108F2}, {0x108F4, 0x108F5}, {0x10900, 0x10915}, {0x10920, 0x10939}, {0x10980, 0x109B7},
+{0x109BE, 0x109BF}, {0x10A00, 0x10A00}, {0x10A10, 0x10A13}, {0x10A15, 0x10A17}, {0x10A19, 0x10A35}, {0x10A60, 0x10A7C}, {0x10A80, 0x10A9C}, {0x10AC0, 0x10AC7}, {0x10AC9, 0x10AE4}, {0x10B00, 0x10B35},
+{0x10B40, 0x10B55}, {0x10B60, 0x10B72}, {0x10B80, 0x10B91}, {0x10C00, 0x10C48}, {0x10C80, 0x10CB2}, {0x10CC0, 0x10CF2}, {0x10D00, 0x10D23}, {0x10E80, 0x10EA9}, {0x10EB0, 0x10EB1}, {0x10F00, 0x10F1C},
+{0x10F27, 0x10F27}, {0x10F30, 0x10F45}, {0x10FB0, 0x10FC4}, {0x10FE0, 0x10FF6}, {0x11003, 0x11037}, {0x11083, 0x110AF}, {0x110D0, 0x110E8}, {0x11103, 0x11126}, {0x11144, 0x11144}, {0x11147, 0x11147},
+{0x11150, 0x11172}, {0x11176, 0x11176}, {0x11183, 0x111B2}, {0x111C1, 0x111C4}, {0x111DA, 0x111DA}, {0x111DC, 0x111DC}, {0x11200, 0x11211}, {0x11213, 0x1122B}, {0x11280, 0x11286}, {0x11288, 0x11288},
+{0x1128A, 0x1128D}, {0x1128F, 0x1129D}, {0x1129F, 0x112A8}, {0x112B0, 0x112DE}, {0x11305, 0x1130C}, {0x1130F, 0x11310}, {0x11313, 0x11328}, {0x1132A, 0x11330}, {0x11332, 0x11333}, {0x11335, 0x11339},
+{0x1133D, 0x1133D}, {0x11350, 0x11350}, {0x1135D, 0x11361}, {0x11400, 0x11434}, {0x11447, 0x1144A}, {0x1145F, 0x11461}, {0x11480, 0x114AF}, {0x114C4, 0x114C5}, {0x114C7, 0x114C7}, {0x11580, 0x115AE},
+{0x115D8, 0x115DB}, {0x11600, 0x1162F}, {0x11644, 0x11644}, {0x11680, 0x116AA}, {0x116B8, 0x116B8}, {0x11700, 0x1171A}, {0x11800, 0x1182B}, {0x118A0, 0x118DF}, {0x118FF, 0x11906}, {0x11909, 0x11909},
+{0x1190C, 0x11913}, {0x11915, 0x11916}, {0x11918, 0x1192F}, {0x1193F, 0x1193F}, {0x11941, 0x11941}, {0x119A0, 0x119A7}, {0x119AA, 0x119D0}, {0x119E1, 0x119E1}, {0x119E3, 0x119E3}, {0x11A00, 0x11A00},
+{0x11A0B, 0x11A32}, {0x11A3A, 0x11A3A}, {0x11A50, 0x11A50}, {0x11A5C, 0x11A89}, {0x11A9D, 0x11A9D}, {0x11AC0, 0x11AF8}, {0x11C00, 0x11C08}, {0x11C0A, 0x11C2E}, {0x11C40, 0x11C40}, {0x11C72, 0x11C8F},
+{0x11D00, 0x11D06}, {0x11D08, 0x11D09}, {0x11D0B, 0x11D30}, {0x11D46, 0x11D46}, {0x11D60, 0x11D65}, {0x11D67, 0x11D68}, {0x11D6A, 0x11D89}, {0x11D98, 0x11D98}, {0x11EE0, 0x11EF2}, {0x11FB0, 0x11FB0},
+{0x12000, 0x12399}, {0x12480, 0x12543}, {0x13000, 0x1342E}, {0x14400, 0x14646}, {0x16800, 0x16A38}, {0x16A40, 0x16A5E}, {0x16AD0, 0x16AED}, {0x16B00, 0x16B2F}, {0x16B40, 0x16B43}, {0x16B63, 0x16B77},
+{0x16B7D, 0x16B8F}, {0x16E40, 0x16E7F}, {0x16F00, 0x16F4A}, {0x16F50, 0x16F50}, {0x16F93, 0x16F9F}, {0x16FE0, 0x16FE1}, {0x16FE3, 0x16FE3}, {0x17000, 0x187F7}, {0x18800, 0x18CD5}, {0x18D00, 0x18D08},
+{0x1B000, 0x1B11E}, {0x1B150, 0x1B152}, {0x1B164, 0x1B167}, {0x1B170, 0x1B2FB}, {0x1BC00, 0x1BC6A}, {0x1BC70, 0x1BC7C}, {0x1BC80, 0x1BC88}, {0x1BC90, 0x1BC99}, {0x1D400, 0x1D454}, {0x1D456, 0x1D49C},
+{0x1D49E, 0x1D49F}, {0x1D4A2, 0x1D4A2}, {0x1D4A5, 0x1D4A6}, {0x1D4A9, 0x1D4AC}, {0x1D4AE, 0x1D4B9}, {0x1D4BB, 0x1D4BB}, {0x1D4BD, 0x1D4C3}, {0x1D4C5, 0x1D505}, {0x1D507, 0x1D50A}, {0x1D50D, 0x1D514},
+{0x1D516, 0x1D51C}, {0x1D51E, 0x1D539}, {0x1D53B, 0x1D53E}, {0x1D540, 0x1D544}, {0x1D546, 0x1D546}, {0x1D54A, 0x1D550}, {0x1D552, 0x1D6A5}, {0x1D6A8, 0x1D6C0}, {0x1D6C2, 0x1D6DA}, {0x1D6DC, 0x1D6FA},
+{0x1D6FC, 0x1D714}, {0x1D716, 0x1D734}, {0x1D736, 0x1D74E}, {0x1D750, 0x1D76E}, {0x1D770, 0x1D788}, {0x1D78A, 0x1D7A8}, {0x1D7AA, 0x1D7C2}, {0x1D7C4, 0x1D7CB}, {0x1E100, 0x1E12C}, {0x1E137, 0x1E13D},
+{0x1E14E, 0x1E14E}, {0x1E2C0, 0x1E2EB}, {0x1E800, 0x1E8C4}, {0x1E900, 0x1E943}, {0x1E94B, 0x1E94B}, {0x1EE00, 0x1EE03}, {0x1EE05, 0x1EE1F}, {0x1EE21, 0x1EE22}, {0x1EE24, 0x1EE24}, {0x1EE27, 0x1EE27},
+{0x1EE29, 0x1EE32}, {0x1EE34, 0x1EE37}, {0x1EE39, 0x1EE39}, {0x1EE3B, 0x1EE3B}, {0x1EE42, 0x1EE42}, {0x1EE47, 0x1EE47}, {0x1EE49, 0x1EE49}, {0x1EE4B, 0x1EE4B}, {0x1EE4D, 0x1EE4F}, {0x1EE51, 0x1EE52},
+{0x1EE54, 0x1EE54}, {0x1EE57, 0x1EE57}, {0x1EE59, 0x1EE59}, {0x1EE5B, 0x1EE5B}, {0x1EE5D, 0x1EE5D}, {0x1EE5F, 0x1EE5F}, {0x1EE61, 0x1EE62}, {0x1EE64, 0x1EE64}, {0x1EE67, 0x1EE6A}, {0x1EE6C, 0x1EE72},
+{0x1EE74, 0x1EE77}, {0x1EE79, 0x1EE7C}, {0x1EE7E, 0x1EE7E}, {0x1EE80, 0x1EE89}, {0x1EE8B, 0x1EE9B}, {0x1EEA1, 0x1EEA3}, {0x1EEA5, 0x1EEA9}, {0x1EEAB, 0x1EEBB}, {0x20000, 0x2A6DD}, {0x2A700, 0x2B734},
+{0x2B740, 0x2B81D}, {0x2B820, 0x2CEA1}, {0x2CEB0, 0x2EBE0}, {0x2F800, 0x2FA1D}, {0x30000, 0x3134A},
+};
+
+static const std::vector<std::pair<uint32_t, uint32_t>> whitespace_ranges = {
+{0x9, 0xD}, {0x1C, 0x20}, {0x85, 0x85}, {0xA0, 0xA0}, {0x1680, 0x1680}, {0x2000, 0x200A}, {0x2028, 0x2029}, {0x202F, 0x202F}, {0x205F, 0x205F}, {0x3000, 0x3000},
+};
+
+static const std::vector<std::pair<uint32_t, uint32_t>> accent_mark_ranges = {
+{0x300, 0x36F}, {0x483, 0x489}, {0x591, 0x5BD}, {0x5BF, 0x5BF}, {0x5C1, 0x5C2}, {0x5C4, 0x5C5}, {0x5C7, 0x5C7}, {0x610, 0x61A}, {0x64B, 0x65F}, {0x670, 0x670}, {0x6D6, 0x6DC}, {0x6DF, 0x6E4},
+{0x6E7, 0x6E8}, {0x6EA, 0x6ED}, {0x711, 0x711}, {0x730, 0x74A}, {0x7A6, 0x7B0}, {0x7EB, 0x7F3}, {0x7FD, 0x7FD}, {0x816, 0x819}, {0x81B, 0x823}, {0x825, 0x827}, {0x829, 0x82D}, {0x859, 0x85B},
+{0x8D3, 0x8E1}, {0x8E3, 0x903}, {0x93A, 0x93C}, {0x93E, 0x94F}, {0x951, 0x957}, {0x962, 0x963}, {0x981, 0x983}, {0x9BC, 0x9BC}, {0x9BE, 0x9C4}, {0x9C7, 0x9C8}, {0x9CB, 0x9CD}, {0x9D7, 0x9D7},
+{0x9E2, 0x9E3}, {0x9FE, 0x9FE}, {0xA01, 0xA03}, {0xA3C, 0xA3C}, {0xA3E, 0xA42}, {0xA47, 0xA48}, {0xA4B, 0xA4D}, {0xA51, 0xA51}, {0xA70, 0xA71}, {0xA75, 0xA75}, {0xA81, 0xA83}, {0xABC, 0xABC},
+{0xABE, 0xAC5}, {0xAC7, 0xAC9}, {0xACB, 0xACD}, {0xAE2, 0xAE3}, {0xAFA, 0xAFF}, {0xB01, 0xB03}, {0xB3C, 0xB3C}, {0xB3E, 0xB44}, {0xB47, 0xB48}, {0xB4B, 0xB4D}, {0xB55, 0xB57}, {0xB62, 0xB63},
+{0xB82, 0xB82}, {0xBBE, 0xBC2}, {0xBC6, 0xBC8}, {0xBCA, 0xBCD}, {0xBD7, 0xBD7}, {0xC00, 0xC04}, {0xC3E, 0xC44}, {0xC46, 0xC48}, {0xC4A, 0xC4D}, {0xC55, 0xC56}, {0xC62, 0xC63}, {0xC81, 0xC83},
+{0xCBC, 0xCBC}, {0xCBE, 0xCC4}, {0xCC6, 0xCC8}, {0xCCA, 0xCCD}, {0xCD5, 0xCD6}, {0xCE2, 0xCE3}, {0xD00, 0xD03}, {0xD3B, 0xD3C}, {0xD3E, 0xD44}, {0xD46, 0xD48}, {0xD4A, 0xD4D}, {0xD57, 0xD57},
+{0xD62, 0xD63}, {0xD81, 0xD83}, {0xDCA, 0xDCA}, {0xDCF, 0xDD4}, {0xDD6, 0xDD6}, {0xDD8, 0xDDF}, {0xDF2, 0xDF3}, {0xE31, 0xE31}, {0xE34, 0xE3A}, {0xE47, 0xE4E}, {0xEB1, 0xEB1}, {0xEB4, 0xEBC},
+{0xEC8, 0xECD}, {0xF18, 0xF19}, {0xF35, 0xF35}, {0xF37, 0xF37}, {0xF39, 0xF39}, {0xF3E, 0xF3F}, {0xF71, 0xF84}, {0xF86, 0xF87}, {0xF8D, 0xF97}, {0xF99, 0xFBC}, {0xFC6, 0xFC6}, {0x102B, 0x103E},
+{0x1056, 0x1059}, {0x105E, 0x1060}, {0x1062, 0x1064}, {0x1067, 0x106D}, {0x1071, 0x1074}, {0x1082, 0x108D}, {0x108F, 0x108F}, {0x109A, 0x109D}, {0x135D, 0x135F}, {0x1712, 0x1714}, {0x1732, 0x1734},
+{0x1752, 0x1753}, {0x1772, 0x1773}, {0x17B4, 0x17D3}, {0x17DD, 0x17DD}, {0x180B, 0x180D}, {0x1885, 0x1886}, {0x18A9, 0x18A9}, {0x1920, 0x192B}, {0x1930, 0x193B}, {0x1A17, 0x1A1B}, {0x1A55, 0x1A5E},
+{0x1A60, 0x1A7C}, {0x1A7F, 0x1A7F}, {0x1AB0, 0x1AC0}, {0x1B00, 0x1B04}, {0x1B34, 0x1B44}, {0x1B6B, 0x1B73}, {0x1B80, 0x1B82}, {0x1BA1, 0x1BAD}, {0x1BE6, 0x1BF3}, {0x1C24, 0x1C37}, {0x1CD0, 0x1CD2},
+{0x1CD4, 0x1CE8}, {0x1CED, 0x1CED}, {0x1CF4, 0x1CF4}, {0x1CF7, 0x1CF9}, {0x1DC0, 0x1DF9}, {0x1DFB, 0x1DFF}, {0x20D0, 0x20F0}, {0x2CEF, 0x2CF1}, {0x2D7F, 0x2D7F}, {0x2DE0, 0x2DFF}, {0x302A, 0x302F},
+{0x3099, 0x309A}, {0xA66F, 0xA672}, {0xA674, 0xA67D}, {0xA69E, 0xA69F}, {0xA6F0, 0xA6F1}, {0xA802, 0xA802}, {0xA806, 0xA806}, {0xA80B, 0xA80B}, {0xA823, 0xA827}, {0xA82C, 0xA82C}, {0xA880, 0xA881},
+{0xA8B4, 0xA8C5}, {0xA8E0, 0xA8F1}, {0xA8FF, 0xA8FF}, {0xA926, 0xA92D}, {0xA947, 0xA953}, {0xA980, 0xA983}, {0xA9B3, 0xA9C0}, {0xA9E5, 0xA9E5}, {0xAA29, 0xAA36}, {0xAA43, 0xAA43}, {0xAA4C, 0xAA4D},
+{0xAA7B, 0xAA7D}, {0xAAB0, 0xAAB0}, {0xAAB2, 0xAAB4}, {0xAAB7, 0xAAB8}, {0xAABE, 0xAABF}, {0xAAC1, 0xAAC1}, {0xAAEB, 0xAAEF}, {0xAAF5, 0xAAF6}, {0xABE3, 0xABEA}, {0xABEC, 0xABED}, {0xFB1E, 0xFB1E},
+{0xFE00, 0xFE0F}, {0xFE20, 0xFE2F}, {0x101FD, 0x101FD}, {0x102E0, 0x102E0}, {0x10376, 0x1037A}, {0x10A01, 0x10A03}, {0x10A05, 0x10A06}, {0x10A0C, 0x10A0F}, {0x10A38, 0x10A3A}, {0x10A3F, 0x10A3F},
+{0x10AE5, 0x10AE6}, {0x10D24, 0x10D27}, {0x10EAB, 0x10EAC}, {0x10F46, 0x10F50}, {0x11000, 0x11002}, {0x11038, 0x11046}, {0x1107F, 0x11082}, {0x110B0, 0x110BA}, {0x11100, 0x11102}, {0x11127, 0x11134},
+{0x11145, 0x11146}, {0x11173, 0x11173}, {0x11180, 0x11182}, {0x111B3, 0x111C0}, {0x111C9, 0x111CC}, {0x111CE, 0x111CF}, {0x1122C, 0x11237}, {0x1123E, 0x1123E}, {0x112DF, 0x112EA}, {0x11300, 0x11303},
+{0x1133B, 0x1133C}, {0x1133E, 0x11344}, {0x11347, 0x11348}, {0x1134B, 0x1134D}, {0x11357, 0x11357}, {0x11362, 0x11363}, {0x11366, 0x1136C}, {0x11370, 0x11374}, {0x11435, 0x11446}, {0x1145E, 0x1145E},
+{0x114B0, 0x114C3}, {0x115AF, 0x115B5}, {0x115B8, 0x115C0}, {0x115DC, 0x115DD}, {0x11630, 0x11640}, {0x116AB, 0x116B7}, {0x1171D, 0x1172B}, {0x1182C, 0x1183A}, {0x11930, 0x11935}, {0x11937, 0x11938},
+{0x1193B, 0x1193E}, {0x11940, 0x11940}, {0x11942, 0x11943}, {0x119D1, 0x119D7}, {0x119DA, 0x119E0}, {0x119E4, 0x119E4}, {0x11A01, 0x11A0A}, {0x11A33, 0x11A39}, {0x11A3B, 0x11A3E}, {0x11A47, 0x11A47},
+{0x11A51, 0x11A5B}, {0x11A8A, 0x11A99}, {0x11C2F, 0x11C36}, {0x11C38, 0x11C3F}, {0x11C92, 0x11CA7}, {0x11CA9, 0x11CB6}, {0x11D31, 0x11D36}, {0x11D3A, 0x11D3A}, {0x11D3C, 0x11D3D}, {0x11D3F, 0x11D45},
+{0x11D47, 0x11D47}, {0x11D8A, 0x11D8E}, {0x11D90, 0x11D91}, {0x11D93, 0x11D97}, {0x11EF3, 0x11EF6}, {0x16AF0, 0x16AF4}, {0x16B30, 0x16B36}, {0x16F4F, 0x16F4F}, {0x16F51, 0x16F87}, {0x16F8F, 0x16F92},
+{0x16FE4, 0x16FE4}, {0x16FF0, 0x16FF1}, {0x1BC9D, 0x1BC9E}, {0x1D165, 0x1D169}, {0x1D16D, 0x1D172}, {0x1D17B, 0x1D182}, {0x1D185, 0x1D18B}, {0x1D1AA, 0x1D1AD}, {0x1D242, 0x1D244}, {0x1DA00, 0x1DA36},
+{0x1DA3B, 0x1DA6C}, {0x1DA75, 0x1DA75}, {0x1DA84, 0x1DA84}, {0x1DA9B, 0x1DA9F}, {0x1DAA1, 0x1DAAF}, {0x1E000, 0x1E006}, {0x1E008, 0x1E018}, {0x1E01B, 0x1E021}, {0x1E023, 0x1E024}, {0x1E026, 0x1E02A},
+{0x1E130, 0x1E136}, {0x1E2EC, 0x1E2EF}, {0x1E8D0, 0x1E8D6}, {0x1E944, 0x1E94A}, {0xE0100, 0xE01EF},
+};
+
+static const std::vector<std::pair<uint32_t, uint32_t>> punctuation_ranges = {
+{0x21, 0x23}, {0x25, 0x2A}, {0x2C, 0x2F}, {0x3A, 0x3B}, {0x3F, 0x40}, {0x5B, 0x5D}, {0x5F, 0x5F}, {0x7B, 0x7B}, {0x7D, 0x7D}, {0xA1, 0xA1}, {0xA7, 0xA7}, {0xAB, 0xAB}, {0xB6, 0xB7}, {0xBB, 0xBB},
+{0xBF, 0xBF}, {0x37E, 0x37E}, {0x387, 0x387}, {0x55A, 0x55F}, {0x589, 0x58A}, {0x5BE, 0x5BE}, {0x5C0, 0x5C0}, {0x5C3, 0x5C3}, {0x5C6, 0x5C6}, {0x5F3, 0x5F4}, {0x609, 0x60A}, {0x60C, 0x60D},
+{0x61B, 0x61B}, {0x61E, 0x61F}, {0x66A, 0x66D}, {0x6D4, 0x6D4}, {0x700, 0x70D}, {0x7F7, 0x7F9}, {0x830, 0x83E}, {0x85E, 0x85E}, {0x964, 0x965}, {0x970, 0x970}, {0x9FD, 0x9FD}, {0xA76, 0xA76},
+{0xAF0, 0xAF0}, {0xC77, 0xC77}, {0xC84, 0xC84}, {0xDF4, 0xDF4}, {0xE4F, 0xE4F}, {0xE5A, 0xE5B}, {0xF04, 0xF12}, {0xF14, 0xF14}, {0xF3A, 0xF3D}, {0xF85, 0xF85}, {0xFD0, 0xFD4}, {0xFD9, 0xFDA},
+{0x104A, 0x104F}, {0x10FB, 0x10FB}, {0x1360, 0x1368}, {0x1400, 0x1400}, {0x166E, 0x166E}, {0x169B, 0x169C}, {0x16EB, 0x16ED}, {0x1735, 0x1736}, {0x17D4, 0x17D6}, {0x17D8, 0x17DA}, {0x1800, 0x180A},
+{0x1944, 0x1945}, {0x1A1E, 0x1A1F}, {0x1AA0, 0x1AA6}, {0x1AA8, 0x1AAD}, {0x1B5A, 0x1B60}, {0x1BFC, 0x1BFF}, {0x1C3B, 0x1C3F}, {0x1C7E, 0x1C7F}, {0x1CC0, 0x1CC7}, {0x1CD3, 0x1CD3}, {0x2010, 0x2027},
+{0x2030, 0x2043}, {0x2045, 0x2051}, {0x2053, 0x205E}, {0x207D, 0x207E}, {0x208D, 0x208E}, {0x2308, 0x230B}, {0x2329, 0x232A}, {0x2768, 0x2775}, {0x27C5, 0x27C6}, {0x27E6, 0x27EF}, {0x2983, 0x2998},
+{0x29D8, 0x29DB}, {0x29FC, 0x29FD}, {0x2CF9, 0x2CFC}, {0x2CFE, 0x2CFF}, {0x2D70, 0x2D70}, {0x2E00, 0x2E2E}, {0x2E30, 0x2E4F}, {0x2E52, 0x2E52}, {0x3001, 0x3003}, {0x3008, 0x3011}, {0x3014, 0x301F},
+{0x3030, 0x3030}, {0x303D, 0x303D}, {0x30A0, 0x30A0}, {0x30FB, 0x30FB}, {0xA4FE, 0xA4FF}, {0xA60D, 0xA60F}, {0xA673, 0xA673}, {0xA67E, 0xA67E}, {0xA6F2, 0xA6F7}, {0xA874, 0xA877}, {0xA8CE, 0xA8CF},
+{0xA8F8, 0xA8FA}, {0xA8FC, 0xA8FC}, {0xA92E, 0xA92F}, {0xA95F, 0xA95F}, {0xA9C1, 0xA9CD}, {0xA9DE, 0xA9DF}, {0xAA5C, 0xAA5F}, {0xAADE, 0xAADF}, {0xAAF0, 0xAAF1}, {0xABEB, 0xABEB}, {0xFD3E, 0xFD3F},
+{0xFE10, 0xFE19}, {0xFE30, 0xFE52}, {0xFE54, 0xFE61}, {0xFE63, 0xFE63}, {0xFE68, 0xFE68}, {0xFE6A, 0xFE6B}, {0xFF01, 0xFF03}, {0xFF05, 0xFF0A}, {0xFF0C, 0xFF0F}, {0xFF1A, 0xFF1B}, {0xFF1F, 0xFF20},
+{0xFF3B, 0xFF3D}, {0xFF3F, 0xFF3F}, {0xFF5B, 0xFF5B}, {0xFF5D, 0xFF5D}, {0xFF5F, 0xFF65}, {0x10100, 0x10102}, {0x1039F, 0x1039F}, {0x103D0, 0x103D0}, {0x1056F, 0x1056F}, {0x10857, 0x10857},
+{0x1091F, 0x1091F}, {0x1093F, 0x1093F}, {0x10A50, 0x10A58}, {0x10A7F, 0x10A7F}, {0x10AF0, 0x10AF6}, {0x10B39, 0x10B3F}, {0x10B99, 0x10B9C}, {0x10EAD, 0x10EAD}, {0x10F55, 0x10F59}, {0x11047, 0x1104D},
+{0x110BB, 0x110BC}, {0x110BE, 0x110C1}, {0x11140, 0x11143}, {0x11174, 0x11175}, {0x111C5, 0x111C8}, {0x111CD, 0x111CD}, {0x111DB, 0x111DB}, {0x111DD, 0x111DF}, {0x11238, 0x1123D}, {0x112A9, 0x112A9},
+{0x1144B, 0x1144F}, {0x1145A, 0x1145B}, {0x1145D, 0x1145D}, {0x114C6, 0x114C6}, {0x115C1, 0x115D7}, {0x11641, 0x11643}, {0x11660, 0x1166C}, {0x1173C, 0x1173E}, {0x1183B, 0x1183B}, {0x11944, 0x11946},
+{0x119E2, 0x119E2}, {0x11A3F, 0x11A46}, {0x11A9A, 0x11A9C}, {0x11A9E, 0x11AA2}, {0x11C41, 0x11C45}, {0x11C70, 0x11C71}, {0x11EF7, 0x11EF8}, {0x11FFF, 0x11FFF}, {0x12470, 0x12474}, {0x16A6E, 0x16A6F},
+{0x16AF5, 0x16AF5}, {0x16B37, 0x16B3B}, {0x16B44, 0x16B44}, {0x16E97, 0x16E9A}, {0x16FE2, 0x16FE2}, {0x1BC9F, 0x1BC9F}, {0x1DA87, 0x1DA8B}, {0x1E95E, 0x1E95F},
+};
+
+static const std::vector<std::pair<uint32_t, uint32_t>> symbol_ranges = {
+{0x24, 0x24}, {0x2B, 0x2B}, {0x3C, 0x3E}, {0x5E, 0x5E}, {0x60, 0x60}, {0x7C, 0x7C}, {0x7E, 0x7E}, {0xA2, 0xA6}, {0xA8, 0xA9}, {0xAC, 0xAC}, {0xAE, 0xB1}, {0xB4, 0xB4}, {0xB8, 0xB8}, {0xD7, 0xD7},
+{0xF7, 0xF7}, {0x2C2, 0x2C5}, {0x2D2, 0x2DF}, {0x2E5, 0x2EB}, {0x2ED, 0x2ED}, {0x2EF, 0x2FF}, {0x375, 0x375}, {0x384, 0x385}, {0x3F6, 0x3F6}, {0x482, 0x482}, {0x58D, 0x58F}, {0x606, 0x608},
+{0x60B, 0x60B}, {0x60E, 0x60F}, {0x6DE, 0x6DE}, {0x6E9, 0x6E9}, {0x6FD, 0x6FE}, {0x7F6, 0x7F6}, {0x7FE, 0x7FF}, {0x9F2, 0x9F3}, {0x9FA, 0x9FB}, {0xAF1, 0xAF1}, {0xB70, 0xB70}, {0xBF3, 0xBFA},
+{0xC7F, 0xC7F}, {0xD4F, 0xD4F}, {0xD79, 0xD79}, {0xE3F, 0xE3F}, {0xF01, 0xF03}, {0xF13, 0xF13}, {0xF15, 0xF17}, {0xF1A, 0xF1F}, {0xF34, 0xF34}, {0xF36, 0xF36}, {0xF38, 0xF38}, {0xFBE, 0xFC5},
+{0xFC7, 0xFCC}, {0xFCE, 0xFCF}, {0xFD5, 0xFD8}, {0x109E, 0x109F}, {0x1390, 0x1399}, {0x166D, 0x166D}, {0x17DB, 0x17DB}, {0x1940, 0x1940}, {0x19DE, 0x19FF}, {0x1B61, 0x1B6A}, {0x1B74, 0x1B7C},
+{0x1FBD, 0x1FBD}, {0x1FBF, 0x1FC1}, {0x1FCD, 0x1FCF}, {0x1FDD, 0x1FDF}, {0x1FED, 0x1FEF}, {0x1FFD, 0x1FFE}, {0x2044, 0x2044}, {0x2052, 0x2052}, {0x207A, 0x207C}, {0x208A, 0x208C}, {0x20A0, 0x20BF},
+{0x2100, 0x2101}, {0x2103, 0x2106}, {0x2108, 0x2109}, {0x2114, 0x2114}, {0x2116, 0x2118}, {0x211E, 0x2123}, {0x2125, 0x2125}, {0x2127, 0x2127}, {0x2129, 0x2129}, {0x212E, 0x212E}, {0x213A, 0x213B},
+{0x2140, 0x2144}, {0x214A, 0x214D}, {0x214F, 0x214F}, {0x218A, 0x218B}, {0x2190, 0x2307}, {0x230C, 0x2328}, {0x232B, 0x2426}, {0x2440, 0x244A}, {0x249C, 0x24E9}, {0x2500, 0x2767}, {0x2794, 0x27C4},
+{0x27C7, 0x27E5}, {0x27F0, 0x2982}, {0x2999, 0x29D7}, {0x29DC, 0x29FB}, {0x29FE, 0x2B73}, {0x2B76, 0x2B95}, {0x2B97, 0x2BFF}, {0x2CE5, 0x2CEA}, {0x2E50, 0x2E51}, {0x2E80, 0x2E99}, {0x2E9B, 0x2EF3},
+{0x2F00, 0x2FD5}, {0x2FF0, 0x2FFB}, {0x3004, 0x3004}, {0x3012, 0x3013}, {0x3020, 0x3020}, {0x3036, 0x3037}, {0x303E, 0x303F}, {0x309B, 0x309C}, {0x3190, 0x3191}, {0x3196, 0x319F}, {0x31C0, 0x31E3},
+{0x3200, 0x321E}, {0x322A, 0x3247}, {0x3250, 0x3250}, {0x3260, 0x327F}, {0x328A, 0x32B0}, {0x32C0, 0x33FF}, {0x4DC0, 0x4DFF}, {0xA490, 0xA4C6}, {0xA700, 0xA716}, {0xA720, 0xA721}, {0xA789, 0xA78A},
+{0xA828, 0xA82B}, {0xA836, 0xA839}, {0xAA77, 0xAA79}, {0xAB5B, 0xAB5B}, {0xAB6A, 0xAB6B}, {0xFB29, 0xFB29}, {0xFBB2, 0xFBC1}, {0xFDFC, 0xFDFD}, {0xFE62, 0xFE62}, {0xFE64, 0xFE66}, {0xFE69, 0xFE69},
+{0xFF04, 0xFF04}, {0xFF0B, 0xFF0B}, {0xFF1C, 0xFF1E}, {0xFF3E, 0xFF3E}, {0xFF40, 0xFF40}, {0xFF5C, 0xFF5C}, {0xFF5E, 0xFF5E}, {0xFFE0, 0xFFE6}, {0xFFE8, 0xFFEE}, {0xFFFC, 0xFFFD}, {0x10137, 0x1013F},
+{0x10179, 0x10189}, {0x1018C, 0x1018E}, {0x10190, 0x1019C}, {0x101A0, 0x101A0}, {0x101D0, 0x101FC}, {0x10877, 0x10878}, {0x10AC8, 0x10AC8}, {0x1173F, 0x1173F}, {0x11FD5, 0x11FF1}, {0x16B3C, 0x16B3F},
+{0x16B45, 0x16B45}, {0x1BC9C, 0x1BC9C}, {0x1D000, 0x1D0F5}, {0x1D100, 0x1D126}, {0x1D129, 0x1D164}, {0x1D16A, 0x1D16C}, {0x1D183, 0x1D184}, {0x1D18C, 0x1D1A9}, {0x1D1AE, 0x1D1E8}, {0x1D200, 0x1D241},
+{0x1D245, 0x1D245}, {0x1D300, 0x1D356}, {0x1D6C1, 0x1D6C1}, {0x1D6DB, 0x1D6DB}, {0x1D6FB, 0x1D6FB}, {0x1D715, 0x1D715}, {0x1D735, 0x1D735}, {0x1D74F, 0x1D74F}, {0x1D76F, 0x1D76F}, {0x1D789, 0x1D789},
+{0x1D7A9, 0x1D7A9}, {0x1D7C3, 0x1D7C3}, {0x1D800, 0x1D9FF}, {0x1DA37, 0x1DA3A}, {0x1DA6D, 0x1DA74}, {0x1DA76, 0x1DA83}, {0x1DA85, 0x1DA86}, {0x1E14F, 0x1E14F}, {0x1E2FF, 0x1E2FF}, {0x1ECAC, 0x1ECAC},
+{0x1ECB0, 0x1ECB0}, {0x1ED2E, 0x1ED2E}, {0x1EEF0, 0x1EEF1}, {0x1F000, 0x1F02B}, {0x1F030, 0x1F093}, {0x1F0A0, 0x1F0AE}, {0x1F0B1, 0x1F0BF}, {0x1F0C1, 0x1F0CF}, {0x1F0D1, 0x1F0F5}, {0x1F10D, 0x1F1AD},
+{0x1F1E6, 0x1F202}, {0x1F210, 0x1F23B}, {0x1F240, 0x1F248}, {0x1F250, 0x1F251}, {0x1F260, 0x1F265}, {0x1F300, 0x1F6D7}, {0x1F6E0, 0x1F6EC}, {0x1F6F0, 0x1F6FC}, {0x1F700, 0x1F773}, {0x1F780, 0x1F7D8},
+{0x1F7E0, 0x1F7EB}, {0x1F800, 0x1F80B}, {0x1F810, 0x1F847}, {0x1F850, 0x1F859}, {0x1F860, 0x1F887}, {0x1F890, 0x1F8AD}, {0x1F8B0, 0x1F8B1}, {0x1F900, 0x1F978}, {0x1F97A, 0x1F9CB}, {0x1F9CD, 0x1FA53},
+{0x1FA60, 0x1FA6D}, {0x1FA70, 0x1FA74}, {0x1FA78, 0x1FA7A}, {0x1FA80, 0x1FA86}, {0x1FA90, 0x1FAA8}, {0x1FAB0, 0x1FAB6}, {0x1FAC0, 0x1FAC2}, {0x1FAD0, 0x1FAD6}, {0x1FB00, 0x1FB92}, {0x1FB94, 0x1FBCA},
+};
+
+static const std::vector<std::pair<uint32_t, uint32_t>> control_ranges = {
+{0x0, 0x8}, {0xE, 0x1B}, {0x7F, 0x84}, {0x86, 0x9F}, {0xAD, 0xAD}, {0x378, 0x379}, {0x380, 0x383}, {0x38B, 0x38B}, {0x38D, 0x38D}, {0x3A2, 0x3A2}, {0x530, 0x530}, {0x557, 0x558}, {0x58B, 0x58C},
+{0x590, 0x590}, {0x5C8, 0x5CF}, {0x5EB, 0x5EE}, {0x5F5, 0x605}, {0x61C, 0x61D}, {0x6DD, 0x6DD}, {0x70E, 0x70F}, {0x74B, 0x74C}, {0x7B2, 0x7BF}, {0x7FB, 0x7FC}, {0x82E, 0x82F}, {0x83F, 0x83F},
+{0x85C, 0x85D}, {0x85F, 0x85F}, {0x86B, 0x89F}, {0x8B5, 0x8B5}, {0x8C8, 0x8D2}, {0x8E2, 0x8E2}, {0x984, 0x984}, {0x98D, 0x98E}, {0x991, 0x992}, {0x9A9, 0x9A9}, {0x9B1, 0x9B1}, {0x9B3, 0x9B5},
+{0x9BA, 0x9BB}, {0x9C5, 0x9C6}, {0x9C9, 0x9CA}, {0x9CF, 0x9D6}, {0x9D8, 0x9DB}, {0x9DE, 0x9DE}, {0x9E4, 0x9E5}, {0x9FF, 0xA00}, {0xA04, 0xA04}, {0xA0B, 0xA0E}, {0xA11, 0xA12}, {0xA29, 0xA29},
+{0xA31, 0xA31}, {0xA34, 0xA34}, {0xA37, 0xA37}, {0xA3A, 0xA3B}, {0xA3D, 0xA3D}, {0xA43, 0xA46}, {0xA49, 0xA4A}, {0xA4E, 0xA50}, {0xA52, 0xA58}, {0xA5D, 0xA5D}, {0xA5F, 0xA65}, {0xA77, 0xA80},
+{0xA84, 0xA84}, {0xA8E, 0xA8E}, {0xA92, 0xA92}, {0xAA9, 0xAA9}, {0xAB1, 0xAB1}, {0xAB4, 0xAB4}, {0xABA, 0xABB}, {0xAC6, 0xAC6}, {0xACA, 0xACA}, {0xACE, 0xACF}, {0xAD1, 0xADF}, {0xAE4, 0xAE5},
+{0xAF2, 0xAF8}, {0xB00, 0xB00}, {0xB04, 0xB04}, {0xB0D, 0xB0E}, {0xB11, 0xB12}, {0xB29, 0xB29}, {0xB31, 0xB31}, {0xB34, 0xB34}, {0xB3A, 0xB3B}, {0xB45, 0xB46}, {0xB49, 0xB4A}, {0xB4E, 0xB54},
+{0xB58, 0xB5B}, {0xB5E, 0xB5E}, {0xB64, 0xB65}, {0xB78, 0xB81}, {0xB84, 0xB84}, {0xB8B, 0xB8D}, {0xB91, 0xB91}, {0xB96, 0xB98}, {0xB9B, 0xB9B}, {0xB9D, 0xB9D}, {0xBA0, 0xBA2}, {0xBA5, 0xBA7},
+{0xBAB, 0xBAD}, {0xBBA, 0xBBD}, {0xBC3, 0xBC5}, {0xBC9, 0xBC9}, {0xBCE, 0xBCF}, {0xBD1, 0xBD6}, {0xBD8, 0xBE5}, {0xBFB, 0xBFF}, {0xC0D, 0xC0D}, {0xC11, 0xC11}, {0xC29, 0xC29}, {0xC3A, 0xC3C},
+{0xC45, 0xC45}, {0xC49, 0xC49}, {0xC4E, 0xC54}, {0xC57, 0xC57}, {0xC5B, 0xC5F}, {0xC64, 0xC65}, {0xC70, 0xC76}, {0xC8D, 0xC8D}, {0xC91, 0xC91}, {0xCA9, 0xCA9}, {0xCB4, 0xCB4}, {0xCBA, 0xCBB},
+{0xCC5, 0xCC5}, {0xCC9, 0xCC9}, {0xCCE, 0xCD4}, {0xCD7, 0xCDD}, {0xCDF, 0xCDF}, {0xCE4, 0xCE5}, {0xCF0, 0xCF0}, {0xCF3, 0xCFF}, {0xD0D, 0xD0D}, {0xD11, 0xD11}, {0xD45, 0xD45}, {0xD49, 0xD49},
+{0xD50, 0xD53}, {0xD64, 0xD65}, {0xD80, 0xD80}, {0xD84, 0xD84}, {0xD97, 0xD99}, {0xDB2, 0xDB2}, {0xDBC, 0xDBC}, {0xDBE, 0xDBF}, {0xDC7, 0xDC9}, {0xDCB, 0xDCE}, {0xDD5, 0xDD5}, {0xDD7, 0xDD7},
+{0xDE0, 0xDE5}, {0xDF0, 0xDF1}, {0xDF5, 0xE00}, {0xE3B, 0xE3E}, {0xE5C, 0xE80}, {0xE83, 0xE83}, {0xE85, 0xE85}, {0xE8B, 0xE8B}, {0xEA4, 0xEA4}, {0xEA6, 0xEA6}, {0xEBE, 0xEBF}, {0xEC5, 0xEC5},
+{0xEC7, 0xEC7}, {0xECE, 0xECF}, {0xEDA, 0xEDB}, {0xEE0, 0xEFF}, {0xF48, 0xF48}, {0xF6D, 0xF70}, {0xF98, 0xF98}, {0xFBD, 0xFBD}, {0xFCD, 0xFCD}, {0xFDB, 0xFFF}, {0x10C6, 0x10C6}, {0x10C8, 0x10CC},
+{0x10CE, 0x10CF}, {0x1249, 0x1249}, {0x124E, 0x124F}, {0x1257, 0x1257}, {0x1259, 0x1259}, {0x125E, 0x125F}, {0x1289, 0x1289}, {0x128E, 0x128F}, {0x12B1, 0x12B1}, {0x12B6, 0x12B7}, {0x12BF, 0x12BF},
+{0x12C1, 0x12C1}, {0x12C6, 0x12C7}, {0x12D7, 0x12D7}, {0x1311, 0x1311}, {0x1316, 0x1317}, {0x135B, 0x135C}, {0x137D, 0x137F}, {0x139A, 0x139F}, {0x13F6, 0x13F7}, {0x13FE, 0x13FF}, {0x169D, 0x169F},
+{0x16F9, 0x16FF}, {0x170D, 0x170D}, {0x1715, 0x171F}, {0x1737, 0x173F}, {0x1754, 0x175F}, {0x176D, 0x176D}, {0x1771, 0x1771}, {0x1774, 0x177F}, {0x17DE, 0x17DF}, {0x17EA, 0x17EF}, {0x17FA, 0x17FF},
+{0x180E, 0x180F}, {0x181A, 0x181F}, {0x1879, 0x187F}, {0x18AB, 0x18AF}, {0x18F6, 0x18FF}, {0x191F, 0x191F}, {0x192C, 0x192F}, {0x193C, 0x193F}, {0x1941, 0x1943}, {0x196E, 0x196F}, {0x1975, 0x197F},
+{0x19AC, 0x19AF}, {0x19CA, 0x19CF}, {0x19DB, 0x19DD}, {0x1A1C, 0x1A1D}, {0x1A5F, 0x1A5F}, {0x1A7D, 0x1A7E}, {0x1A8A, 0x1A8F}, {0x1A9A, 0x1A9F}, {0x1AAE, 0x1AAF}, {0x1AC1, 0x1AFF}, {0x1B4C, 0x1B4F},
+{0x1B7D, 0x1B7F}, {0x1BF4, 0x1BFB}, {0x1C38, 0x1C3A}, {0x1C4A, 0x1C4C}, {0x1C89, 0x1C8F}, {0x1CBB, 0x1CBC}, {0x1CC8, 0x1CCF}, {0x1CFB, 0x1CFF}, {0x1DFA, 0x1DFA}, {0x1F16, 0x1F17}, {0x1F1E, 0x1F1F},
+{0x1F46, 0x1F47}, {0x1F4E, 0x1F4F}, {0x1F58, 0x1F58}, {0x1F5A, 0x1F5A}, {0x1F5C, 0x1F5C}, {0x1F5E, 0x1F5E}, {0x1F7E, 0x1F7F}, {0x1FB5, 0x1FB5}, {0x1FC5, 0x1FC5}, {0x1FD4, 0x1FD5}, {0x1FDC, 0x1FDC},
+{0x1FF0, 0x1FF1}, {0x1FF5, 0x1FF5}, {0x1FFF, 0x1FFF}, {0x200B, 0x200F}, {0x202A, 0x202E}, {0x2060, 0x206F}, {0x2072, 0x2073}, {0x208F, 0x208F}, {0x209D, 0x209F}, {0x20C0, 0x20CF}, {0x20F1, 0x20FF},
+{0x218C, 0x218F}, {0x2427, 0x243F}, {0x244B, 0x245F}, {0x2B74, 0x2B75}, {0x2B96, 0x2B96}, {0x2C2F, 0x2C2F}, {0x2C5F, 0x2C5F}, {0x2CF4, 0x2CF8}, {0x2D26, 0x2D26}, {0x2D28, 0x2D2C}, {0x2D2E, 0x2D2F},
+{0x2D68, 0x2D6E}, {0x2D71, 0x2D7E}, {0x2D97, 0x2D9F}, {0x2DA7, 0x2DA7}, {0x2DAF, 0x2DAF}, {0x2DB7, 0x2DB7}, {0x2DBF, 0x2DBF}, {0x2DC7, 0x2DC7}, {0x2DCF, 0x2DCF}, {0x2DD7, 0x2DD7}, {0x2DDF, 0x2DDF},
+{0x2E53, 0x2E7F}, {0x2E9A, 0x2E9A}, {0x2EF4, 0x2EFF}, {0x2FD6, 0x2FEF}, {0x2FFC, 0x2FFF}, {0x3040, 0x3040}, {0x3097, 0x3098}, {0x3100, 0x3104}, {0x3130, 0x3130}, {0x318F, 0x318F}, {0x31E4, 0x31EF},
+{0x321F, 0x321F}, {0x9FFD, 0x9FFF}, {0xA48D, 0xA48F}, {0xA4C7, 0xA4CF}, {0xA62C, 0xA63F}, {0xA6F8, 0xA6FF}, {0xA7C0, 0xA7C1}, {0xA7CB, 0xA7F4}, {0xA82D, 0xA82F}, {0xA83A, 0xA83F}, {0xA878, 0xA87F},
+{0xA8C6, 0xA8CD}, {0xA8DA, 0xA8DF}, {0xA954, 0xA95E}, {0xA97D, 0xA97F}, {0xA9CE, 0xA9CE}, {0xA9DA, 0xA9DD}, {0xA9FF, 0xA9FF}, {0xAA37, 0xAA3F}, {0xAA4E, 0xAA4F}, {0xAA5A, 0xAA5B}, {0xAAC3, 0xAADA},
+{0xAAF7, 0xAB00}, {0xAB07, 0xAB08}, {0xAB0F, 0xAB10}, {0xAB17, 0xAB1F}, {0xAB27, 0xAB27}, {0xAB2F, 0xAB2F}, {0xAB6C, 0xAB6F}, {0xABEE, 0xABEF}, {0xABFA, 0xABFF}, {0xD7A4, 0xD7AF}, {0xD7C7, 0xD7CA},
+{0xD7FC, 0xF8FF}, {0xFA6E, 0xFA6F}, {0xFADA, 0xFAFF}, {0xFB07, 0xFB12}, {0xFB18, 0xFB1C}, {0xFB37, 0xFB37}, {0xFB3D, 0xFB3D}, {0xFB3F, 0xFB3F}, {0xFB42, 0xFB42}, {0xFB45, 0xFB45}, {0xFBC2, 0xFBD2},
+{0xFD40, 0xFD4F}, {0xFD90, 0xFD91}, {0xFDC8, 0xFDEF}, {0xFDFE, 0xFDFF}, {0xFE1A, 0xFE1F}, {0xFE53, 0xFE53}, {0xFE67, 0xFE67}, {0xFE6C, 0xFE6F}, {0xFE75, 0xFE75}, {0xFEFD, 0xFF00}, {0xFFBF, 0xFFC1},
+{0xFFC8, 0xFFC9}, {0xFFD0, 0xFFD1}, {0xFFD8, 0xFFD9}, {0xFFDD, 0xFFDF}, {0xFFE7, 0xFFE7}, {0xFFEF, 0xFFFB}, {0xFFFE, 0xFFFF}, {0x1000C, 0x1000C}, {0x10027, 0x10027}, {0x1003B, 0x1003B},
+{0x1003E, 0x1003E}, {0x1004E, 0x1004F}, {0x1005E, 0x1007F}, {0x100FB, 0x100FF}, {0x10103, 0x10106}, {0x10134, 0x10136}, {0x1018F, 0x1018F}, {0x1019D, 0x1019F}, {0x101A1, 0x101CF}, {0x101FE, 0x1027F},
+{0x1029D, 0x1029F}, {0x102D1, 0x102DF}, {0x102FC, 0x102FF}, {0x10324, 0x1032C}, {0x1034B, 0x1034F}, {0x1037B, 0x1037F}, {0x1039E, 0x1039E}, {0x103C4, 0x103C7}, {0x103D6, 0x103FF}, {0x1049E, 0x1049F},
+{0x104AA, 0x104AF}, {0x104D4, 0x104D7}, {0x104FC, 0x104FF}, {0x10528, 0x1052F}, {0x10564, 0x1056E}, {0x10570, 0x105FF}, {0x10737, 0x1073F}, {0x10756, 0x1075F}, {0x10768, 0x107FF}, {0x10806, 0x10807},
+{0x10809, 0x10809}, {0x10836, 0x10836}, {0x10839, 0x1083B}, {0x1083D, 0x1083E}, {0x10856, 0x10856}, {0x1089F, 0x108A6}, {0x108B0, 0x108DF}, {0x108F3, 0x108F3}, {0x108F6, 0x108FA}, {0x1091C, 0x1091E},
+{0x1093A, 0x1093E}, {0x10940, 0x1097F}, {0x109B8, 0x109BB}, {0x109D0, 0x109D1}, {0x10A04, 0x10A04}, {0x10A07, 0x10A0B}, {0x10A14, 0x10A14}, {0x10A18, 0x10A18}, {0x10A36, 0x10A37}, {0x10A3B, 0x10A3E},
+{0x10A49, 0x10A4F}, {0x10A59, 0x10A5F}, {0x10AA0, 0x10ABF}, {0x10AE7, 0x10AEA}, {0x10AF7, 0x10AFF}, {0x10B36, 0x10B38}, {0x10B56, 0x10B57}, {0x10B73, 0x10B77}, {0x10B92, 0x10B98}, {0x10B9D, 0x10BA8},
+{0x10BB0, 0x10BFF}, {0x10C49, 0x10C7F}, {0x10CB3, 0x10CBF}, {0x10CF3, 0x10CF9}, {0x10D28, 0x10D2F}, {0x10D3A, 0x10E5F}, {0x10E7F, 0x10E7F}, {0x10EAA, 0x10EAA}, {0x10EAE, 0x10EAF}, {0x10EB2, 0x10EFF},
+{0x10F28, 0x10F2F}, {0x10F5A, 0x10FAF}, {0x10FCC, 0x10FDF}, {0x10FF7, 0x10FFF}, {0x1104E, 0x11051}, {0x11070, 0x1107E}, {0x110BD, 0x110BD}, {0x110C2, 0x110CF}, {0x110E9, 0x110EF}, {0x110FA, 0x110FF},
+{0x11135, 0x11135}, {0x11148, 0x1114F}, {0x11177, 0x1117F}, {0x111E0, 0x111E0}, {0x111F5, 0x111FF}, {0x11212, 0x11212}, {0x1123F, 0x1127F}, {0x11287, 0x11287}, {0x11289, 0x11289}, {0x1128E, 0x1128E},
+{0x1129E, 0x1129E}, {0x112AA, 0x112AF}, {0x112EB, 0x112EF}, {0x112FA, 0x112FF}, {0x11304, 0x11304}, {0x1130D, 0x1130E}, {0x11311, 0x11312}, {0x11329, 0x11329}, {0x11331, 0x11331}, {0x11334, 0x11334},
+{0x1133A, 0x1133A}, {0x11345, 0x11346}, {0x11349, 0x1134A}, {0x1134E, 0x1134F}, {0x11351, 0x11356}, {0x11358, 0x1135C}, {0x11364, 0x11365}, {0x1136D, 0x1136F}, {0x11375, 0x113FF}, {0x1145C, 0x1145C},
+{0x11462, 0x1147F}, {0x114C8, 0x114CF}, {0x114DA, 0x1157F}, {0x115B6, 0x115B7}, {0x115DE, 0x115FF}, {0x11645, 0x1164F}, {0x1165A, 0x1165F}, {0x1166D, 0x1167F}, {0x116B9, 0x116BF}, {0x116CA, 0x116FF},
+{0x1171B, 0x1171C}, {0x1172C, 0x1172F}, {0x11740, 0x117FF}, {0x1183C, 0x1189F}, {0x118F3, 0x118FE}, {0x11907, 0x11908}, {0x1190A, 0x1190B}, {0x11914, 0x11914}, {0x11917, 0x11917}, {0x11936, 0x11936},
+{0x11939, 0x1193A}, {0x11947, 0x1194F}, {0x1195A, 0x1199F}, {0x119A8, 0x119A9}, {0x119D8, 0x119D9}, {0x119E5, 0x119FF}, {0x11A48, 0x11A4F}, {0x11AA3, 0x11ABF}, {0x11AF9, 0x11BFF}, {0x11C09, 0x11C09},
+{0x11C37, 0x11C37}, {0x11C46, 0x11C4F}, {0x11C6D, 0x11C6F}, {0x11C90, 0x11C91}, {0x11CA8, 0x11CA8}, {0x11CB7, 0x11CFF}, {0x11D07, 0x11D07}, {0x11D0A, 0x11D0A}, {0x11D37, 0x11D39}, {0x11D3B, 0x11D3B},
+{0x11D3E, 0x11D3E}, {0x11D48, 0x11D4F}, {0x11D5A, 0x11D5F}, {0x11D66, 0x11D66}, {0x11D69, 0x11D69}, {0x11D8F, 0x11D8F}, {0x11D92, 0x11D92}, {0x11D99, 0x11D9F}, {0x11DAA, 0x11EDF}, {0x11EF9, 0x11FAF},
+{0x11FB1, 0x11FBF}, {0x11FF2, 0x11FFE}, {0x1239A, 0x123FF}, {0x1246F, 0x1246F}, {0x12475, 0x1247F}, {0x12544, 0x12FFF}, {0x1342F, 0x143FF}, {0x14647, 0x167FF}, {0x16A39, 0x16A3F}, {0x16A5F, 0x16A5F},
+{0x16A6A, 0x16A6D}, {0x16A70, 0x16ACF}, {0x16AEE, 0x16AEF}, {0x16AF6, 0x16AFF}, {0x16B46, 0x16B4F}, {0x16B5A, 0x16B5A}, {0x16B62, 0x16B62}, {0x16B78, 0x16B7C}, {0x16B90, 0x16E3F}, {0x16E9B, 0x16EFF},
+{0x16F4B, 0x16F4E}, {0x16F88, 0x16F8E}, {0x16FA0, 0x16FDF}, {0x16FE5, 0x16FEF}, {0x16FF2, 0x16FFF}, {0x187F8, 0x187FF}, {0x18CD6, 0x18CFF}, {0x18D09, 0x1AFFF}, {0x1B11F, 0x1B14F}, {0x1B153, 0x1B163},
+{0x1B168, 0x1B16F}, {0x1B2FC, 0x1BBFF}, {0x1BC6B, 0x1BC6F}, {0x1BC7D, 0x1BC7F}, {0x1BC89, 0x1BC8F}, {0x1BC9A, 0x1BC9B}, {0x1BCA0, 0x1CFFF}, {0x1D0F6, 0x1D0FF}, {0x1D127, 0x1D128}, {0x1D173, 0x1D17A},
+{0x1D1E9, 0x1D1FF}, {0x1D246, 0x1D2DF}, {0x1D2F4, 0x1D2FF}, {0x1D357, 0x1D35F}, {0x1D379, 0x1D3FF}, {0x1D455, 0x1D455}, {0x1D49D, 0x1D49D}, {0x1D4A0, 0x1D4A1}, {0x1D4A3, 0x1D4A4}, {0x1D4A7, 0x1D4A8},
+{0x1D4AD, 0x1D4AD}, {0x1D4BA, 0x1D4BA}, {0x1D4BC, 0x1D4BC}, {0x1D4C4, 0x1D4C4}, {0x1D506, 0x1D506}, {0x1D50B, 0x1D50C}, {0x1D515, 0x1D515}, {0x1D51D, 0x1D51D}, {0x1D53A, 0x1D53A}, {0x1D53F, 0x1D53F},
+{0x1D545, 0x1D545}, {0x1D547, 0x1D549}, {0x1D551, 0x1D551}, {0x1D6A6, 0x1D6A7}, {0x1D7CC, 0x1D7CD}, {0x1DA8C, 0x1DA9A}, {0x1DAA0, 0x1DAA0}, {0x1DAB0, 0x1DFFF}, {0x1E007, 0x1E007}, {0x1E019, 0x1E01A},
+{0x1E022, 0x1E022}, {0x1E025, 0x1E025}, {0x1E02B, 0x1E0FF}, {0x1E12D, 0x1E12F}, {0x1E13E, 0x1E13F}, {0x1E14A, 0x1E14D}, {0x1E150, 0x1E2BF}, {0x1E2FA, 0x1E2FE}, {0x1E300, 0x1E7FF}, {0x1E8C5, 0x1E8C6},
+{0x1E8D7, 0x1E8FF}, {0x1E94C, 0x1E94F}, {0x1E95A, 0x1E95D}, {0x1E960, 0x1EC70}, {0x1ECB5, 0x1ED00}, {0x1ED3E, 0x1EDFF}, {0x1EE04, 0x1EE04}, {0x1EE20, 0x1EE20}, {0x1EE23, 0x1EE23}, {0x1EE25, 0x1EE26},
+{0x1EE28, 0x1EE28}, {0x1EE33, 0x1EE33}, {0x1EE38, 0x1EE38}, {0x1EE3A, 0x1EE3A}, {0x1EE3C, 0x1EE41}, {0x1EE43, 0x1EE46}, {0x1EE48, 0x1EE48}, {0x1EE4A, 0x1EE4A}, {0x1EE4C, 0x1EE4C}, {0x1EE50, 0x1EE50},
+{0x1EE53, 0x1EE53}, {0x1EE55, 0x1EE56}, {0x1EE58, 0x1EE58}, {0x1EE5A, 0x1EE5A}, {0x1EE5C, 0x1EE5C}, {0x1EE5E, 0x1EE5E}, {0x1EE60, 0x1EE60}, {0x1EE63, 0x1EE63}, {0x1EE65, 0x1EE66}, {0x1EE6B, 0x1EE6B},
+{0x1EE73, 0x1EE73}, {0x1EE78, 0x1EE78}, {0x1EE7D, 0x1EE7D}, {0x1EE7F, 0x1EE7F}, {0x1EE8A, 0x1EE8A}, {0x1EE9C, 0x1EEA0}, {0x1EEA4, 0x1EEA4}, {0x1EEAA, 0x1EEAA}, {0x1EEBC, 0x1EEEF}, {0x1EEF2, 0x1EFFF},
+{0x1F02C, 0x1F02F}, {0x1F094, 0x1F09F}, {0x1F0AF, 0x1F0B0}, {0x1F0C0, 0x1F0C0}, {0x1F0D0, 0x1F0D0}, {0x1F0F6, 0x1F0FF}, {0x1F1AE, 0x1F1E5}, {0x1F203, 0x1F20F}, {0x1F23C, 0x1F23F}, {0x1F249, 0x1F24F},
+{0x1F252, 0x1F25F}, {0x1F266, 0x1F2FF}, {0x1F6D8, 0x1F6DF}, {0x1F6ED, 0x1F6EF}, {0x1F6FD, 0x1F6FF}, {0x1F774, 0x1F77F}, {0x1F7D9, 0x1F7DF}, {0x1F7EC, 0x1F7FF}, {0x1F80C, 0x1F80F}, {0x1F848, 0x1F84F},
+{0x1F85A, 0x1F85F}, {0x1F888, 0x1F88F}, {0x1F8AE, 0x1F8AF}, {0x1F8B2, 0x1F8FF}, {0x1F979, 0x1F979}, {0x1F9CC, 0x1F9CC}, {0x1FA54, 0x1FA5F}, {0x1FA6E, 0x1FA6F}, {0x1FA75, 0x1FA77}, {0x1FA7B, 0x1FA7F},
+{0x1FA87, 0x1FA8F}, {0x1FAA9, 0x1FAAF}, {0x1FAB7, 0x1FABF}, {0x1FAC3, 0x1FACF}, {0x1FAD7, 0x1FAFF}, {0x1FB93, 0x1FB93}, {0x1FBCB, 0x1FBEF}, {0x1FBFA, 0x1FFFF}, {0x2A6DE, 0x2A6FF}, {0x2B735, 0x2B73F},
+{0x2B81E, 0x2B81F}, {0x2CEA2, 0x2CEAF}, {0x2EBE1, 0x2F7FF}, {0x2FA1E, 0x2FFFF}, {0x3134B, 0xE00FF}, {0xE01F0, 0x10FFFF},
+};
+
+static std::string codepoint_to_utf8(uint32_t cp) {
+    std::string result;
+    if (/* 0x00 <= cp && */ cp <= 0x7f) {
+        result.push_back(cp);
+    }
+    else if (0x80 <= cp && cp <= 0x7ff) {
+        result.push_back(0xc0 | ((cp >> 6) & 0x1f));
+        result.push_back(0x80 | (cp & 0x3f));
+    }
+    else if (0x800 <= cp && cp <= 0xffff) {
+        result.push_back(0xe0 | ((cp >> 12) & 0x0f));
+        result.push_back(0x80 | ((cp >> 6) & 0x3f));
+        result.push_back(0x80 | (cp & 0x3f));
+    }
+    else if (0x10000 <= cp && cp <= 0x10ffff) {
+        result.push_back(0xf0 | ((cp >> 18) & 0x07));
+        result.push_back(0x80 | ((cp >> 12) & 0x3f));
+        result.push_back(0x80 | ((cp >> 6) & 0x3f));
+        result.push_back(0x80 | (cp & 0x3f));
+    }
+    else {
+        throw std::invalid_argument("invalid codepoint");
+    }
+    return result;
+}
+
+static std::string codepoints_to_utf8(const std::vector<uint32_t> & cps) {
+    std::string result;
+    for (size_t i = 0; i < cps.size(); ++i) {
+        result.append(codepoint_to_utf8(cps[i]));
+    }
+    return result;
+}
+
+static uint32_t codepoint_from_utf8(const std::string & utf8, size_t & offset) {
+    assert(offset < utf8.size());
+    if (!(utf8[offset + 0] & 0x80)) {
+        auto result = utf8[offset + 0];
+        offset += 1;
+        return result;
+    }
+    else if (!(utf8[offset + 0] & 0x40)) {
+        throw std::invalid_argument("invalid character");
+    }
+    else if (!(utf8[offset + 0] & 0x20)) {
+        if (offset + 1 >= utf8.size() || ! ((utf8[offset + 1] & 0xc0) == 0x80))
+            throw std::invalid_argument("invalid character");
+        auto result = ((utf8[offset + 0] & 0x1f) << 6) | (utf8[offset + 1] & 0x3f);
+        offset += 2;
+        return result;
+    }
+    else if (!(utf8[offset + 0] & 0x10)) {
+        if (offset + 2 >= utf8.size() || ! ((utf8[offset + 1] & 0xc0) == 0x80) || ! ((utf8[offset + 2] & 0xc0) == 0x80))
+            throw std::invalid_argument("invalid character");
+        auto result = ((utf8[offset + 0] & 0x0f) << 12) | ((utf8[offset + 1] & 0x3f) << 6) | (utf8[offset + 2] & 0x3f);
+        offset += 3;
+        return result;
+    }
+    else if (!(utf8[offset + 0] & 0x08)) {
+        if (offset + 3 >= utf8.size() || ! ((utf8[offset + 1] & 0xc0) == 0x80) || ! ((utf8[offset + 2] & 0xc0) == 0x80) || !((utf8[offset + 3] & 0xc0) == 0x80))
+            throw std::invalid_argument("invalid character");
+        auto result = ((utf8[offset + 0] & 0x07) << 18) | ((utf8[offset + 1] & 0x3f) << 12) | ((utf8[offset + 2] & 0x3f) << 6) | (utf8[offset + 3] & 0x3f);
+        offset += 4;
+        return result;
+    }
+    throw std::invalid_argument("invalid string");
+}
+
+static std::vector<uint32_t> codepoints_from_utf8(const std::string & utf8) {
+    std::vector<uint32_t> result;
+    size_t offset = 0;
+    while (offset < utf8.size()) {
+        result.push_back(codepoint_from_utf8(utf8, offset));
+    }
+    return result;
+}
+
+static std::vector<uint16_t> codepoint_to_utf16(uint32_t cp) {
+    std::vector<uint16_t> result;
+    if (/* 0x0000 <= cp && */ cp <= 0xffff) {
+        result.emplace_back(cp);
+    }
+    else if (0x10000 <= cp && cp <= 0x10ffff) {
+        result.emplace_back(0xd800 | ((cp - 0x10000) >> 10));
+        result.emplace_back(0xdc00 | ((cp - 0x10000) & 0x03ff));
+    }
+    else {
+        throw std::invalid_argument("invalid codepoint");
+    }
+    return result;
+}
+
+static std::vector<uint16_t> codepoints_to_utf16(const std::vector<uint32_t> & cps) {
+    std::vector<uint16_t> result;
+    for (size_t i = 0; i < cps.size(); ++i) {
+        auto temp = codepoint_to_utf16(cps[i]);
+        result.insert(result.end(), temp.begin(), temp.end());
+    }
+    return result;
+}
+
+static uint32_t codepoint_from_utf16(const std::vector<uint16_t> & utf16, size_t & offset) {
+    assert(offset < utf16.size());
+    if (((utf16[0] >> 10) << 10) != 0xd800) {
+        auto result = utf16[offset + 0];
+        offset += 1;
+        return result;
+    }
+    else {
+        if (offset + 1 >= utf16.size() || !((utf16[1] & 0xdc00) == 0xdc00))
+            throw std::invalid_argument("invalid character");
+        auto result = 0x10000 + (((utf16[0] & 0x03ff) << 10) | (utf16[1] & 0x03ff));
+        offset += 2;
+        return result;
+    }
+    throw std::invalid_argument("invalid string");
+}
+
+static std::vector<uint32_t> codepoints_from_utf16(const std::vector<uint16_t> & utf16) {
+    std::vector<uint32_t> result;
+    size_t offset = 0;
+    while (offset < utf16.size())
+        result.push_back(codepoint_from_utf16(utf16, offset));
+    return result;
+}
+
+#define CODEPOINT_TYPE_UNIDENTIFIED 0
+#define CODEPOINT_TYPE_DIGIT 1
+#define CODEPOINT_TYPE_LETTER 2
+#define CODEPOINT_TYPE_WHITESPACE 3
+#define CODEPOINT_TYPE_ACCENT_MARK 4
+#define CODEPOINT_TYPE_PUNCTUATION 5
+#define CODEPOINT_TYPE_SYMBOL 6
+#define CODEPOINT_TYPE_CONTROL 7
+
+static std::unordered_map<uint32_t, int> codepoint_type_map() {
+    std::unordered_map<uint32_t, int> codepoint_types;
+    for (auto p : digit_ranges) {
+        for(auto i = p.first; i <= p.second; ++ i)
+            codepoint_types[i] = CODEPOINT_TYPE_DIGIT;
+    }
+    for(auto p : letter_ranges) {
+        for(auto i = p.first; i <= p.second; ++ i)
+            codepoint_types[i] = CODEPOINT_TYPE_LETTER;
+    }
+    for(auto p : whitespace_ranges) {
+        for(auto i = p.first; i <= p.second; ++ i)
+            codepoint_types[i] = CODEPOINT_TYPE_WHITESPACE;
+    }
+    for(auto p : accent_mark_ranges) {
+        for(auto i = p.first; i <= p.second; ++ i)
+            codepoint_types[i] = CODEPOINT_TYPE_ACCENT_MARK;
+    }
+    for(auto p : punctuation_ranges) {
+        for(auto i = p.first; i <= p.second; ++ i)
+            codepoint_types[i] = CODEPOINT_TYPE_PUNCTUATION;
+    }
+    for (auto p : symbol_ranges) {
+        for (auto i = p.first; i <= p.second; ++i)
+            codepoint_types[i] = CODEPOINT_TYPE_SYMBOL;
+    }
+    for(auto p : control_ranges) {
+        for(auto i = p.first; i <= p.second; ++ i)
+            codepoint_types[i] = CODEPOINT_TYPE_CONTROL;
+    }
+    return codepoint_types;
+}
+
+static int codepoint_type(uint32_t cp) {
+    static std::unordered_map<uint32_t, int> codepoint_types = codepoint_type_map();
+    return codepoint_types[cp];
+}
+
+static int codepoint_type(const std::string & utf8) {
+    if (utf8.length() == 0)
+        return CODEPOINT_TYPE_UNIDENTIFIED;
+    size_t offset = 0;
+    return codepoint_type(codepoint_from_utf8(utf8, offset));
+}
+
+static std::unordered_map<uint8_t, std::string> bytes_to_unicode_map_bpe() {
+    std::unordered_map<uint8_t, std::string> map;
+    for (int ch = u'!'; ch <= u'~'; ++ch) {
+        assert(0 <= ch && ch < 256);
+        map[ch] = codepoint_to_utf8(ch);
+    }
+    for (int ch = u'¡'; ch <= u'¬'; ++ch) {
+        assert(0 <= ch && ch < 256);
+        map[ch] = codepoint_to_utf8(ch);
+    }
+    for (int ch = u'®'; ch <= u'ÿ'; ++ch) {
+        assert(0 <= ch && ch < 256);
+        map[ch] = codepoint_to_utf8(ch);
+    }
+    auto n = 0;
+    for (int ch = 0; ch < 256; ++ch) {
+        if (map.find(ch) == map.end()) {
+            map[ch] = codepoint_to_utf8(256 + n);
+            ++n;
+        }
+    }
+    return map;
+}
+
+static std::string bytes_to_unicode_bpe(uint8_t byte) {
+    static std::unordered_map<uint8_t, std::string> map = bytes_to_unicode_map_bpe();
+    return map.at(byte);
+}
+
+static std::unordered_map<std::string, uint8_t> unicode_to_bytes_map_bpe() {
+    std::unordered_map<std::string, uint8_t> map;
+    for (int ch = u'!'; ch <= u'~'; ++ch) {
+        assert(0 <= ch && ch < 256);
+        map[codepoint_to_utf8(ch)] = ch;
+    }
+    for (int ch = u'¡'; ch <= u'¬'; ++ch) {
+        assert(0 <= ch && ch < 256);
+        map[codepoint_to_utf8(ch)] = ch;
+    }
+    for (int ch = u'®'; ch <= u'ÿ'; ++ch) {
+        assert(0 <= ch && ch < 256);
+        map[codepoint_to_utf8(ch)] = ch;
+    }
+    auto n = 0;
+    for (int ch = 0; ch < 256; ++ch) {
+        if (map.find(codepoint_to_utf8(ch)) == map.end()) {
+            map[codepoint_to_utf8(256 + n)] = ch;
+            ++n;
+        }
+    }
+    return map;
+}
+
+static uint8_t unicode_to_bytes_bpe(const std::string & utf8) {
+    static std::unordered_map<std::string, uint8_t> map = unicode_to_bytes_map_bpe();
+    return map.at(utf8);
+}
+

runner/update.sh

@@ -0,0 +1,70 @@
+#!/bin/sh
+
+set -eu
+
+
+status() { echo >&2 ">>> $*"; }
+error() { status "ERROR $*"; }
+usage() {
+    echo "usage: $(basename $0) /path/to/repo"
+    exit 1
+}
+
+OUT=$(dirname $0)
+while getopts "hC:" OPTION; do
+    case $OPTION in
+        C) OUT=$OPTARG ;;
+        *) usage ;;
+    esac
+done
+
+shift $(( $OPTIND - 1 ))
+[ $# -eq 1 ] || usage
+
+status "updating source..."
+cp -a "$1"/*.{c,h,cpp,m,metal,cu} "$OUT"
+
+status "removing incompatible files..."
+rm -f "$OUT"/build-info.h
+
+SHA1=$(git -C $1 rev-parse @)
+
+LICENSE=$(mktemp)
+cleanup() {
+    rm -f $LICENSE
+}
+trap cleanup 0
+
+cat <<EOF | sed 's/ *$//' >$LICENSE
+/**
+ * llama.cpp - git $SHA1
+ *
+$(sed 's/^/ * /' <$1/LICENSE)
+ */
+
+EOF
+
+for IN in $OUT/*.{c,h,cpp,m,metal,cu}; do
+    TMP=$(mktemp)
+    status "updating license $IN"
+    cat $LICENSE $IN >$TMP
+    mv $TMP $IN
+done
+
+touchup() {
+    local CONSTRAINT=$1 && shift
+
+    for IN in $*; do
+        status "touching up $IN..."
+        TMP=$(mktemp)
+        {
+            echo "//go:build $CONSTRAINT"
+            echo
+        } | cat - $IN >$TMP
+        mv $TMP $IN
+    done
+}
+
+touchup darwin $OUT/ggml-metal.*
+touchup mpi $OUT/ggml-mpi.*
+touchup opencl $OUT/ggml-opencl.*

scripts/install.sh

@@ -26,7 +26,8 @@ require() {
 
 [ "$(uname -s)" = "Linux" ] || error 'This script is intended to run on Linux only.'
 
-case "$(uname -m)" in
+ARCH=$(uname -m)
+case "$ARCH" in
     x86_64) ARCH="amd64" ;;
     aarch64|arm64) ARCH="arm64" ;;
     *) error "Unsupported architecture: $ARCH" ;;

scripts/push_docker.sh

@@ -0,0 +1,15 @@
+#!/bin/sh
+
+set -eu
+
+export VERSION=${VERSION:-0.0.0}
+export GOFLAGS="'-ldflags=-w -s \"-X=github.com/jmorganca/ollama/version.Version=$VERSION\" \"-X=github.com/jmorganca/ollama/server.mode=release\"'"
+
+docker buildx build \
+    --push \
+    --platform=linux/arm64,linux/amd64 \
+    --build-arg=VERSION \
+    --build-arg=GOFLAGS \
+    -f Dockerfile \
+    -t ollama/ollama -t ollama/ollama:$VERSION \
+    .

server/auth.go

@@ -106,6 +106,7 @@ func getAuthToken(ctx context.Context, redirData AuthRedirect) (string, error) {
 	resp, err := makeRequest(ctx, "GET", redirectURL, headers, nil, nil)
 	if err != nil {
 		log.Printf("couldn't get token: %q", err)
+		return "", err
 	}
 	defer resp.Body.Close()
 

server/images.go

@@ -1,7 +1,6 @@
 package server
 
 import (
-	"bufio"
 	"bytes"
 	"context"
 	"crypto/sha256"
@@ -26,7 +25,6 @@ import (
 	"github.com/jmorganca/ollama/api"
 	"github.com/jmorganca/ollama/llm"
 	"github.com/jmorganca/ollama/parser"
-	"github.com/jmorganca/ollama/vector"
 	"github.com/jmorganca/ollama/version"
 )
 
@@ -47,12 +45,10 @@ type Model struct {
 	System        string
 	License       []string
 	Digest        string
-	ConfigDigest  string
 	Options       map[string]interface{}
-	Embeddings    []vector.Embedding
 }
 
-func (m *Model) Prompt(request api.GenerateRequest, embedding string) (string, error) {
+func (m *Model) Prompt(request api.GenerateRequest) (string, error) {
 	t := m.Template
 	if request.Template != "" {
 		t = request.Template
@@ -67,7 +63,6 @@ func (m *Model) Prompt(request api.GenerateRequest, embedding string) (string, e
 		First  bool
 		System string
 		Prompt string
-		Embed  string
 
 		// deprecated: versions <= 0.0.7 used this to omit the system prompt
 		Context []int
@@ -77,7 +72,6 @@ func (m *Model) Prompt(request api.GenerateRequest, embedding string) (string, e
 	vars.System = m.System
 	vars.Prompt = request.Prompt
 	vars.Context = request.Context
-	vars.Embed = embedding
 
 	if request.System != "" {
 		vars.System = request.System
@@ -171,12 +165,11 @@ func GetModel(name string) (*Model, error) {
 	}
 
 	model := &Model{
-		Name:         mp.GetFullTagname(),
-		ShortName:    mp.GetShortTagname(),
-		Digest:       digest,
-		ConfigDigest: manifest.Config.Digest,
-		Template:     "{{ .Prompt }}",
-		License:      []string{},
+		Name:      mp.GetFullTagname(),
+		ShortName: mp.GetShortTagname(),
+		Digest:    digest,
+		Template:  "{{ .Prompt }}",
+		License:   []string{},
 	}
 
 	for _, layer := range manifest.Layers {
@@ -190,15 +183,9 @@ func GetModel(name string) (*Model, error) {
 			model.ModelPath = filename
 			model.OriginalModel = layer.From
 		case "application/vnd.ollama.image.embed":
-			file, err := os.Open(filename)
-			if err != nil {
-				return nil, fmt.Errorf("failed to open file: %s", filename)
-			}
-			defer file.Close()
-
-			if err = json.NewDecoder(file).Decode(&model.Embeddings); err != nil {
-				return nil, err
-			}
+			// Deprecated in versions  > 0.1.2
+			// TODO: remove this warning in a future version
+			log.Print("WARNING: model contains embeddings, but embeddings in modelfiles have been deprecated and will be ignored.")
 		case "application/vnd.ollama.image.adapter":
 			model.AdapterPaths = append(model.AdapterPaths, filename)
 		case "application/vnd.ollama.image.template":
@@ -265,7 +252,7 @@ func filenameWithPath(path, f string) (string, error) {
 	return f, nil
 }
 
-func CreateModel(ctx context.Context, workDir, name string, path string, fn func(resp api.ProgressResponse)) error {
+func CreateModel(ctx context.Context, name string, path string, fn func(resp api.ProgressResponse)) error {
 	mp := ParseModelPath(name)
 
 	var manifest *ManifestV2
@@ -310,13 +297,11 @@ func CreateModel(ctx context.Context, workDir, name string, path string, fn func
 	var layers []*LayerReader
 	params := make(map[string][]string)
 	var sourceParams map[string]any
-	embed := EmbeddingParams{fn: fn}
 	for _, c := range commands {
 		log.Printf("[%s] - %s\n", c.Name, c.Args)
 		switch c.Name {
 		case "model":
 			fn(api.ProgressResponse{Status: "looking for model"})
-			embed.model = c.Args
 
 			mp := ParseModelPath(c.Args)
 			mf, _, err := GetManifest(mp)
@@ -340,7 +325,6 @@ func CreateModel(ctx context.Context, workDir, name string, path string, fn func
 						return err
 					}
 				} else {
-					embed.model = modelFile
 					// create a model from this specified file
 					fn(api.ProgressResponse{Status: "creating model layer"})
 					file, err := os.Open(modelFile)
@@ -421,12 +405,6 @@ func CreateModel(ctx context.Context, workDir, name string, path string, fn func
 					layers = append(layers, newLayer)
 				}
 			}
-		case "embed":
-			embedFilePath, err := filenameWithPath(path, c.Args)
-			if err != nil {
-				return err
-			}
-			embed.files = append(embed.files, embedFilePath)
 		case "adapter":
 			fn(api.ProgressResponse{Status: fmt.Sprintf("creating model %s layer", c.Name)})
 
@@ -517,18 +495,8 @@ func CreateModel(ctx context.Context, workDir, name string, path string, fn func
 		}
 		l.MediaType = "application/vnd.ollama.image.params"
 		layers = append(layers, l)
-
-		// apply these parameters to the embedding options, in case embeddings need to be generated using this model
-		embed.opts = formattedParams
 	}
 
-	// generate the embedding layers
-	embeddingLayers, err := embeddingLayers(workDir, embed)
-	if err != nil {
-		return err
-	}
-	layers = append(layers, embeddingLayers...)
-
 	digests, err := getLayerDigests(layers)
 	if err != nil {
 		return err
@@ -572,146 +540,6 @@ func CreateModel(ctx context.Context, workDir, name string, path string, fn func
 	return nil
 }
 
-type EmbeddingParams struct {
-	model string
-	opts  map[string]interface{}
-	files []string // paths to files to embed
-	fn    func(resp api.ProgressResponse)
-}
-
-// embeddingLayers loads the associated LLM and generates the embeddings to be stored from an input file
-func embeddingLayers(workDir string, e EmbeddingParams) ([]*LayerReader, error) {
-	layers := []*LayerReader{}
-	if len(e.files) > 0 {
-		// check if the model is a file path or a model name
-		model, err := GetModel(e.model)
-		if err != nil {
-			if !strings.Contains(err.Error(), "couldn't open file") {
-				return nil, fmt.Errorf("unexpected error opening model to generate embeddings: %v", err)
-			}
-			// the model may be a file path, create a model from this file
-			model = &Model{ModelPath: e.model}
-		}
-
-		if err := load(context.Background(), workDir, model, e.opts, defaultSessionDuration); err != nil {
-			return nil, fmt.Errorf("load model to generate embeddings: %v", err)
-		}
-
-		// this will be used to check if we already have embeddings for a file
-		modelInfo, err := os.Stat(model.ModelPath)
-		if err != nil {
-			return nil, fmt.Errorf("failed to get model file info: %v", err)
-		}
-
-		addedFiles := make(map[string]bool) // keep track of files that have already been added
-		for _, filePattern := range e.files {
-			matchingFiles, err := filepath.Glob(filePattern)
-			if err != nil {
-				return nil, fmt.Errorf("could not find files with pattern %s: %w", filePattern, err)
-			}
-
-			for _, filePath := range matchingFiles {
-				if addedFiles[filePath] {
-					continue
-				}
-				addedFiles[filePath] = true
-				// check if we already have embeddings for this file path
-				layerIdentifier := fmt.Sprintf("%s:%s:%s:%d", filePath, e.model, modelInfo.ModTime().Format("2006-01-02 15:04:05"), modelInfo.Size())
-				digest, _ := GetSHA256Digest(strings.NewReader(layerIdentifier))
-				existing, err := existingFileEmbeddings(digest)
-				if err != nil {
-					return nil, fmt.Errorf("failed to check existing embeddings for file %s: %v", filePath, err)
-				}
-
-				// TODO: check file type
-				f, err := os.Open(filePath)
-				if err != nil {
-					return nil, fmt.Errorf("could not open embed file: %w", err)
-				}
-				scanner := bufio.NewScanner(f)
-				scanner.Split(bufio.ScanLines)
-
-				data := []string{}
-				for scanner.Scan() {
-					data = append(data, scanner.Text())
-				}
-				f.Close()
-
-				// the digest of the file is set here so that the client knows a new operation is in progress
-				fileDigest, _ := GetSHA256Digest(bytes.NewReader([]byte(filePath)))
-
-				embeddings := []vector.Embedding{}
-				for i, d := range data {
-					if strings.TrimSpace(d) == "" {
-						continue
-					}
-					e.fn(api.ProgressResponse{
-						Status:    fmt.Sprintf("creating embeddings for file %s", filePath),
-						Digest:    fileDigest,
-						Total:     int64(len(data) - 1),
-						Completed: int64(i),
-					})
-					if len(existing[d]) > 0 {
-						// already have an embedding for this line
-						embeddings = append(embeddings, vector.Embedding{Data: d, Vector: existing[d]})
-						continue
-					}
-					embed, err := loaded.llm.Embedding(context.Background(), d)
-					if err != nil {
-						log.Printf("failed to generate embedding for '%s' line %d: %v", filePath, i+1, err)
-						continue
-					}
-					embeddings = append(embeddings, vector.Embedding{Data: d, Vector: embed})
-				}
-
-				b, err := json.Marshal(embeddings)
-				if err != nil {
-					return nil, fmt.Errorf("failed to encode embeddings: %w", err)
-				}
-				r := bytes.NewReader(b)
-
-				layer := &LayerReader{
-					Layer: Layer{
-						MediaType: "application/vnd.ollama.image.embed",
-						Digest:    digest,
-						Size:      r.Size(),
-					},
-					Reader: r,
-				}
-
-				layers = append(layers, layer)
-			}
-		}
-	}
-	return layers, nil
-}
-
-// existingFileEmbeddings checks if we already have embeddings for a file and loads them into a look-up map
-func existingFileEmbeddings(digest string) (map[string][]float64, error) {
-	path, err := GetBlobsPath(digest)
-	if err != nil {
-		return nil, fmt.Errorf("embeddings blobs path: %w", err)
-	}
-	existingFileEmbeddings := make(map[string][]float64)
-	if _, err := os.Stat(path); err == nil {
-		// already have some embeddings for this file, load embeddings previously generated
-		file, err := os.Open(path)
-		if err != nil {
-			return nil, fmt.Errorf("failed to open existing embedding file: %s", err)
-		}
-		defer file.Close()
-
-		existing := []vector.Embedding{}
-		if err = json.NewDecoder(file).Decode(&existing); err != nil {
-			return nil, err
-		}
-		for _, e := range existing {
-			existingFileEmbeddings[e.Data] = e.Vector
-		}
-	}
-	return existingFileEmbeddings, nil
-}
-
 func removeLayerFromLayers(layers []*LayerReader, mediaType string) []*LayerReader {
 	return slices.DeleteFunc(layers, func(layer *LayerReader) bool {
 		return layer.MediaType == mediaType
@@ -727,8 +555,7 @@ func SaveLayers(layers []*LayerReader, fn func(resp api.ProgressResponse), force
 		}
 
 		_, err = os.Stat(fp)
-		// note: embed layers are always written since their digest doesnt indicate anything about the contents
-		if os.IsNotExist(err) || force || layer.MediaType == "application/vnd.ollama.image.embed" {
+		if os.IsNotExist(err) || force {
 			fn(api.ProgressResponse{Status: fmt.Sprintf("writing layer %s", layer.Digest)})
 
 			out, err := os.Create(fp)
@@ -1069,51 +896,27 @@ func DeleteModel(name string) error {
 }
 
 func ShowModelfile(model *Model) (string, error) {
-	type modelTemplate struct {
+	var mt struct {
 		*Model
-		From   string
-		Params string
+		From       string
+		Parameters map[string][]any
 	}
 
-	var params []string
+	mt.Parameters = make(map[string][]any)
 	for k, v := range model.Options {
-		switch val := v.(type) {
-		case string:
-			params = append(params, fmt.Sprintf("PARAMETER %s %s", k, val))
-		case int:
-			params = append(params, fmt.Sprintf("PARAMETER %s %s", k, strconv.Itoa(val)))
-		case float64:
-			params = append(params, fmt.Sprintf("PARAMETER %s %s", k, strconv.FormatFloat(val, 'f', 0, 64)))
-		case bool:
-			params = append(params, fmt.Sprintf("PARAMETER %s %s", k, strconv.FormatBool(val)))
-		case []interface{}:
-			for _, nv := range val {
-				switch nval := nv.(type) {
-				case string:
-					params = append(params, fmt.Sprintf("PARAMETER %s %s", k, nval))
-				case int:
-					params = append(params, fmt.Sprintf("PARAMETER %s %s", k, strconv.Itoa(nval)))
-				case float64:
-					params = append(params, fmt.Sprintf("PARAMETER %s %s", k, strconv.FormatFloat(nval, 'f', 0, 64)))
-				case bool:
-					params = append(params, fmt.Sprintf("PARAMETER %s %s", k, strconv.FormatBool(nval)))
-				default:
-					log.Printf("unknown type: %s", reflect.TypeOf(nv).String())
-				}
-			}
-		default:
-			log.Printf("unknown type: %s", reflect.TypeOf(v).String())
+		if s, ok := v.([]any); ok {
+			mt.Parameters[k] = s
+			continue
 		}
+
+		mt.Parameters[k] = []any{v}
 	}
 
-	mt := modelTemplate{
-		Model:  model,
-		From:   model.OriginalModel,
-		Params: strings.Join(params, "\n"),
-	}
+	mt.Model = model
+	mt.From = model.ModelPath
 
-	if mt.From == "" {
-		mt.From = model.ModelPath
+	if model.OriginalModel != "" {
+		mt.From = model.OriginalModel
 	}
 
 	modelFile := `# Modelfile generated by "ollama show"
@@ -1122,12 +925,20 @@ func ShowModelfile(model *Model) (string, error) {
 
 FROM {{ .From }}
 TEMPLATE """{{ .Template }}"""
+
+{{- if .System }}
 SYSTEM """{{ .System }}"""
-{{ .Params }}
-`
-	for _, l := range mt.Model.AdapterPaths {
-		modelFile += fmt.Sprintf("ADAPTER %s\n", l)
-	}
+{{- end }}
+
+{{- range $adapter := .AdapterPaths }}
+ADAPTER {{ $adapter }}
+{{- end }}
+
+{{- range $k, $v := .Parameters }}
+{{- range $parameter := $v }}
+PARAMETER {{ $k }} {{ printf "%#v" $parameter }}
+{{- end }}
+{{- end }}`
 
 	tmpl, err := template.New("").Parse(modelFile)
 	if err != nil {

server/images_test.go

@@ -12,7 +12,7 @@ func TestModelPrompt(t *testing.T) {
 		Template: "a{{ .Prompt }}b",
 		Prompt:   "<h1>",
 	}
-	s, err := m.Prompt(req, "")
+	s, err := m.Prompt(req)
 	if err != nil {
 		t.Fatal(err)
 	}

server/routes.go

@@ -23,11 +23,10 @@ import (
 
 	"github.com/gin-contrib/cors"
 	"github.com/gin-gonic/gin"
-	"gonum.org/v1/gonum/mat"
 
 	"github.com/jmorganca/ollama/api"
 	"github.com/jmorganca/ollama/llm"
-	"github.com/jmorganca/ollama/vector"
+	"github.com/jmorganca/ollama/version"
 )
 
 var mode string = gin.DebugMode
@@ -47,14 +46,13 @@ func init() {
 var loaded struct {
 	mu sync.Mutex
 
-	llm        llm.LLM
-	Embeddings []vector.Embedding
+	runner llm.LLM
 
 	expireAt    time.Time
 	expireTimer *time.Timer
 
-	digest  string
-	options api.Options
+	*Model
+	*api.Options
 }
 
 var defaultSessionDuration = 5 * time.Minute
@@ -72,66 +70,52 @@ func load(ctx context.Context, workDir string, model *Model, reqOpts map[string]
 	}
 
 	// check if the loaded model is still running in a subprocess, in case something unexpected happened
-	if loaded.llm != nil {
-		if err := loaded.llm.Ping(ctx); err != nil {
+	if loaded.runner != nil {
+		if err := loaded.runner.Ping(ctx); err != nil {
 			log.Print("loaded llm process not responding, closing now")
 			// the subprocess is no longer running, so close it
-			loaded.llm.Close()
-			loaded.llm = nil
-			loaded.digest = ""
+			loaded.runner.Close()
+			loaded.runner = nil
+			loaded.Model = nil
+			loaded.Options = nil
 		}
 	}
 
-	if model.Digest != loaded.digest || !reflect.DeepEqual(loaded.options, opts) {
-		if loaded.llm != nil {
+	needLoad := loaded.runner == nil || // is there a model loaded?
+		loaded.ModelPath != model.ModelPath || // has the base model changed?
+		!reflect.DeepEqual(loaded.AdapterPaths, model.AdapterPaths) || // have the adapters changed?
+		!reflect.DeepEqual(loaded.Options.Runner, opts.Runner) // have the runner options changed?
+
+	if needLoad {
+		if loaded.runner != nil {
 			log.Println("changing loaded model")
-			loaded.llm.Close()
-			loaded.llm = nil
-			loaded.digest = ""
+			loaded.runner.Close()
+			loaded.runner = nil
+			loaded.Model = nil
+			loaded.Options = nil
 		}
 
-		if model.Embeddings != nil && len(model.Embeddings) > 0 {
-			opts.EmbeddingOnly = true // this is requried to generate embeddings, completions will still work
-			loaded.Embeddings = model.Embeddings
-		}
-
-		llmModel, err := llm.New(workDir, model.ModelPath, model.AdapterPaths, opts)
+		llmRunner, err := llm.New(workDir, model.ModelPath, model.AdapterPaths, opts)
 		if err != nil {
+			// some older models are not compatible with newer versions of llama.cpp
+			// show a generalized compatibility error until there is a better way to
+			// check for model compatibility
+			if strings.Contains(err.Error(), "failed to load model") {
+				err = fmt.Errorf("%v: this model may be incompatible with your version of Ollama. If you previously pulled this model, try updating it by running `ollama pull %s`", err, model.ShortName)
+			}
+
 			return err
 		}
 
-		// set cache values before modifying opts
-		loaded.llm = llmModel
-		loaded.digest = model.Digest
-		loaded.options = opts
-
-		if opts.NumKeep < 0 {
-			promptWithSystem, err := model.Prompt(api.GenerateRequest{}, "")
-			if err != nil {
-				return err
-			}
-
-			promptNoSystem, err := model.Prompt(api.GenerateRequest{Context: []int{0}}, "")
-			if err != nil {
-				return err
-			}
-
-			tokensWithSystem, err := llmModel.Encode(ctx, promptWithSystem)
-			if err != nil {
-				return err
-			}
-
-			tokensNoSystem, err := llmModel.Encode(ctx, promptNoSystem)
-			if err != nil {
-				return err
-			}
-
-			opts.NumKeep = len(tokensWithSystem) - len(tokensNoSystem)
-
-			llmModel.SetOptions(opts)
-		}
+		loaded.Model = model
+		loaded.runner = llmRunner
+		loaded.Options = &opts
 	}
 
+	// update options for the loaded llm
+	// TODO(mxyng): this isn't thread safe, but it should be fine for now
+	loaded.runner.SetOptions(opts)
+
 	loaded.expireAt = time.Now().Add(sessionDuration)
 
 	if loaded.expireTimer == nil {
@@ -143,13 +127,13 @@ func load(ctx context.Context, workDir string, model *Model, reqOpts map[string]
 				return
 			}
 
-			if loaded.llm == nil {
-				return
+			if loaded.runner != nil {
+				loaded.runner.Close()
 			}
 
-			loaded.llm.Close()
-			loaded.llm = nil
-			loaded.digest = ""
+			loaded.runner = nil
+			loaded.Model = nil
+			loaded.Options = nil
 		})
 	}
 
@@ -164,8 +148,18 @@ func GenerateHandler(c *gin.Context) {
 	checkpointStart := time.Now()
 
 	var req api.GenerateRequest
-	if err := c.ShouldBindJSON(&req); err != nil {
-		c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+	err := c.ShouldBindJSON(&req)
+	switch {
+	case errors.Is(err, io.EOF):
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": "missing request body"})
+		return
+	case err != nil:
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+		return
+	}
+
+	if req.Model == "" {
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": "model is required"})
 		return
 	}
 
@@ -195,22 +189,7 @@ func GenerateHandler(c *gin.Context) {
 
 	checkpointLoaded := time.Now()
 
-	embedding := ""
-	if model.Embeddings != nil && len(model.Embeddings) > 0 {
-		promptEmbed, err := loaded.llm.Embedding(c.Request.Context(), req.Prompt)
-		if err != nil {
-			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
-			return
-		}
-		// TODO: set embed_top from specified parameters in modelfile
-		embed_top := 3
-		topK := vector.TopK(embed_top, mat.NewVecDense(len(promptEmbed), promptEmbed), loaded.Embeddings)
-		for _, e := range topK {
-			embedding = fmt.Sprintf("%s %s", embedding, e.Embedding.Data)
-		}
-	}
-
-	prompt, err := model.Prompt(req, embedding)
+	prompt, err := model.Prompt(req)
 	if err != nil {
 		c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
 		return
@@ -219,6 +198,12 @@ func GenerateHandler(c *gin.Context) {
 	ch := make(chan any)
 	go func() {
 		defer close(ch)
+		// an empty request loads the model
+		if req.Prompt == "" && req.Template == "" && req.System == "" {
+			ch <- api.GenerateResponse{CreatedAt: time.Now().UTC(), Model: req.Model, Done: true}
+			return
+		}
+
 		fn := func(r api.GenerateResponse) {
 			loaded.expireAt = time.Now().Add(sessionDuration)
 			loaded.expireTimer.Reset(sessionDuration)
@@ -233,13 +218,8 @@ func GenerateHandler(c *gin.Context) {
 			ch <- r
 		}
 
-		// an empty request loads the model
-		if req.Prompt == "" && req.Template == "" && req.System == "" {
-			ch <- api.GenerateResponse{Model: req.Model, Done: true}
-		} else {
-			if err := loaded.llm.Predict(c.Request.Context(), req.Context, prompt, fn); err != nil {
-				ch <- gin.H{"error": err.Error()}
-			}
+		if err := loaded.runner.Predict(c.Request.Context(), req.Context, prompt, fn); err != nil {
+			ch <- gin.H{"error": err.Error()}
 		}
 	}()
 
@@ -268,8 +248,18 @@ func EmbeddingHandler(c *gin.Context) {
 	defer loaded.mu.Unlock()
 
 	var req api.EmbeddingRequest
-	if err := c.ShouldBindJSON(&req); err != nil {
-		c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+	err := c.ShouldBindJSON(&req)
+	switch {
+	case errors.Is(err, io.EOF):
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": "missing request body"})
+		return
+	case err != nil:
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+		return
+	}
+
+	if req.Model == "" {
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": "model is required"})
 		return
 	}
 
@@ -285,12 +275,12 @@ func EmbeddingHandler(c *gin.Context) {
 		return
 	}
 
-	if !loaded.options.EmbeddingOnly {
+	if !loaded.Options.EmbeddingOnly {
 		c.JSON(http.StatusBadRequest, gin.H{"error": "embedding option must be set to true"})
 		return
 	}
 
-	embedding, err := loaded.llm.Embedding(c.Request.Context(), req.Prompt)
+	embedding, err := loaded.runner.Embedding(c.Request.Context(), req.Prompt)
 	if err != nil {
 		log.Printf("embedding generation failed: %v", err)
 		c.JSON(http.StatusInternalServerError, gin.H{"error": "failed to generate embedding"})
@@ -305,8 +295,18 @@ func EmbeddingHandler(c *gin.Context) {
 
 func PullModelHandler(c *gin.Context) {
 	var req api.PullRequest
-	if err := c.ShouldBindJSON(&req); err != nil {
-		c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+	err := c.ShouldBindJSON(&req)
+	switch {
+	case errors.Is(err, io.EOF):
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": "missing request body"})
+		return
+	case err != nil:
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+		return
+	}
+
+	if req.Name == "" {
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": "name is required"})
 		return
 	}
 
@@ -339,8 +339,18 @@ func PullModelHandler(c *gin.Context) {
 
 func PushModelHandler(c *gin.Context) {
 	var req api.PushRequest
-	if err := c.ShouldBindJSON(&req); err != nil {
-		c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+	err := c.ShouldBindJSON(&req)
+	switch {
+	case errors.Is(err, io.EOF):
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": "missing request body"})
+		return
+	case err != nil:
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+		return
+	}
+
+	if req.Name == "" {
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": "name is required"})
 		return
 	}
 
@@ -371,12 +381,20 @@ func PushModelHandler(c *gin.Context) {
 
 func CreateModelHandler(c *gin.Context) {
 	var req api.CreateRequest
-	if err := c.ShouldBindJSON(&req); err != nil {
-		c.JSON(http.StatusBadRequest, gin.H{"message": err.Error()})
+	err := c.ShouldBindJSON(&req)
+	switch {
+	case errors.Is(err, io.EOF):
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": "missing request body"})
+		return
+	case err != nil:
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": err.Error()})
 		return
 	}
 
-	workDir := c.GetString("workDir")
+	if req.Name == "" || req.Path == "" {
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": "name and path are required"})
+		return
+	}
 
 	ch := make(chan any)
 	go func() {
@@ -388,7 +406,7 @@ func CreateModelHandler(c *gin.Context) {
 		ctx, cancel := context.WithCancel(c.Request.Context())
 		defer cancel()
 
-		if err := CreateModel(ctx, workDir, req.Name, req.Path, fn); err != nil {
+		if err := CreateModel(ctx, req.Name, req.Path, fn); err != nil {
 			ch <- gin.H{"error": err.Error()}
 		}
 	}()
@@ -403,8 +421,18 @@ func CreateModelHandler(c *gin.Context) {
 
 func DeleteModelHandler(c *gin.Context) {
 	var req api.DeleteRequest
-	if err := c.ShouldBindJSON(&req); err != nil {
-		c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+	err := c.ShouldBindJSON(&req)
+	switch {
+	case errors.Is(err, io.EOF):
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": "missing request body"})
+		return
+	case err != nil:
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+		return
+	}
+
+	if req.Name == "" {
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": "name is required"})
 		return
 	}
 
@@ -433,8 +461,18 @@ func DeleteModelHandler(c *gin.Context) {
 
 func ShowModelHandler(c *gin.Context) {
 	var req api.ShowRequest
-	if err := c.ShouldBindJSON(&req); err != nil {
-		c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+	err := c.ShouldBindJSON(&req)
+	switch {
+	case errors.Is(err, io.EOF):
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": "missing request body"})
+		return
+	case err != nil:
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+		return
+	}
+
+	if req.Name == "" {
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": "name is required"})
 		return
 	}
 
@@ -503,7 +541,7 @@ func GetModelInfo(name string) (*api.ShowResponse, error) {
 }
 
 func ListModelsHandler(c *gin.Context) {
-	var models []api.ModelResponse
+	models := make([]api.ModelResponse, 0)
 	fp, err := GetManifestPath()
 	if err != nil {
 		c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
@@ -544,8 +582,18 @@ func ListModelsHandler(c *gin.Context) {
 
 func CopyModelHandler(c *gin.Context) {
 	var req api.CopyRequest
-	if err := c.ShouldBindJSON(&req); err != nil {
-		c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+	err := c.ShouldBindJSON(&req)
+	switch {
+	case errors.Is(err, io.EOF):
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": "missing request body"})
+		return
+	case err != nil:
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+		return
+	}
+
+	if req.Source == "" || req.Destination == "" {
+		c.AbortWithStatusJSON(http.StatusBadRequest, gin.H{"error": "source add destination are required"})
 		return
 	}
 
@@ -611,7 +659,7 @@ func Serve(ln net.Listener, allowOrigins []string) error {
 		r.Handle(method, "/api/tags", ListModelsHandler)
 	}
 
-	log.Printf("Listening on %s", ln.Addr())
+	log.Printf("Listening on %s (version %s)", ln.Addr(), version.Version)
 	s := &http.Server{
 		Handler: r,
 	}
@@ -621,8 +669,8 @@ func Serve(ln net.Listener, allowOrigins []string) error {
 	signal.Notify(signals, syscall.SIGINT, syscall.SIGTERM)
 	go func() {
 		<-signals
-		if loaded.llm != nil {
-			loaded.llm.Close()
+		if loaded.runner != nil {
+			loaded.runner.Close()
 		}
 		os.RemoveAll(workDir)
 		os.Exit(0)

vector/store.go

@@ -1,69 +0,0 @@
-package vector
-
-import (
-	"container/heap"
-	"sort"
-
-	"gonum.org/v1/gonum/mat"
-)
-
-type Embedding struct {
-	Vector []float64 // the embedding vector
-	Data   string    // the data represted by the embedding
-}
-
-type EmbeddingSimilarity struct {
-	Embedding  Embedding // the embedding that was used to calculate the similarity
-	Similarity float64   // the similarity between the embedding and the query
-}
-
-type Heap []EmbeddingSimilarity
-
-func (h Heap) Len() int           { return len(h) }
-func (h Heap) Less(i, j int) bool { return h[i].Similarity < h[j].Similarity }
-func (h Heap) Swap(i, j int)      { h[i], h[j] = h[j], h[i] }
-func (h *Heap) Push(e any) {
-	*h = append(*h, e.(EmbeddingSimilarity))
-}
-
-func (h *Heap) Pop() interface{} {
-	old := *h
-	n := len(old)
-	x := old[n-1]
-	*h = old[0 : n-1]
-	return x
-}
-
-// cosineSimilarity is a measure that calculates the cosine of the angle between two vectors.
-// This value will range from -1 to 1, where 1 means the vectors are identical.
-func cosineSimilarity(vec1, vec2 *mat.VecDense) float64 {
-	dotProduct := mat.Dot(vec1, vec2)
-	norms := mat.Norm(vec1, 2) * mat.Norm(vec2, 2)
-
-	if norms == 0 {
-		return 0
-	}
-	return dotProduct / norms
-}
-
-func TopK(k int, query *mat.VecDense, embeddings []Embedding) []EmbeddingSimilarity {
-	h := &Heap{}
-	heap.Init(h)
-	for _, emb := range embeddings {
-		similarity := cosineSimilarity(query, mat.NewVecDense(len(emb.Vector), emb.Vector))
-		heap.Push(h, EmbeddingSimilarity{Embedding: emb, Similarity: similarity})
-		if h.Len() > k {
-			heap.Pop(h)
-		}
-	}
-
-	topK := make([]EmbeddingSimilarity, 0, h.Len())
-	for h.Len() > 0 {
-		topK = append(topK, heap.Pop(h).(EmbeddingSimilarity))
-	}
-	sort.Slice(topK, func(i, j int) bool {
-		return topK[i].Similarity > topK[j].Similarity
-	})
-
-	return topK
-}