change error handler behavior and fix error when a model isn't found

.dockerignore

@@ -1,8 +1,7 @@
+build
+llama/build
+.venv
 .vscode
 ollama
 app
-dist
-scripts
-llm/llama.cpp/ggml
-llm/llama.cpp/gguf
-.env
+web

.gitignore

@@ -2,7 +2,5 @@
 .vscode
 .env
 .venv
-.swp
 dist
 ollama
-ggml-metal.metal

.gitmodules

@@ -1,10 +0,0 @@
-[submodule "llm/llama.cpp/ggml"]
-    path = llm/llama.cpp/ggml
-    url = https://github.com/ggerganov/llama.cpp.git
-    ignore = dirty
-    shallow = true
-[submodule "llm/llama.cpp/gguf"]
-    path = llm/llama.cpp/gguf
-    url = https://github.com/ggerganov/llama.cpp.git
-    ignore = dirty
-    shallow = true

Dockerfile

@@ -1,23 +1,15 @@
-FROM nvidia/cuda:11.8.0-devel-ubuntu22.04
-
-ARG TARGETARCH
-ARG VERSION=0.0.0
-ARG GOFLAGS="'-ldflags=-w -s'"
-
+FROM golang:1.20
 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-$TARGETARCH.tar.gz /tmp/go1.21.1.tar.gz
-RUN mkdir -p /usr/local && tar xz -C /usr/local </tmp/go1.21.1.tar.gz
-
 COPY . .
-ENV GOARCH=$TARGETARCH
-RUN /usr/local/go/bin/go generate ./... \
-    && /usr/local/go/bin/go build .
+RUN CGO_ENABLED=1 go build -ldflags '-linkmode external -extldflags "-static"' .
 
-FROM ubuntu:22.04
-RUN apt-get update && apt-get install -y ca-certificates
+FROM alpine
 COPY --from=0 /go/src/github.com/jmorganca/ollama/ollama /bin/ollama
 EXPOSE 11434
-ENV OLLAMA_HOST 0.0.0.0
+ARG USER=ollama
+ARG GROUP=ollama
+RUN addgroup -g 1000 $GROUP && adduser -u 1000 -DG $GROUP $USER
+USER $USER:$GROUP
 ENTRYPOINT ["/bin/ollama"]
+ENV OLLAMA_HOST 0.0.0.0
 CMD ["serve"]

Dockerfile.build

@@ -1,32 +0,0 @@
-
-# centos7 amd64 dependencies
-FROM --platform=linux/amd64 nvidia/cuda:11.8.0-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
-RUN wget "https://github.com/Kitware/CMake/releases/download/v3.27.6/cmake-3.27.6-linux-x86_64.sh" -O cmake-installer.sh && chmod +x cmake-installer.sh && ./cmake-installer.sh --skip-license --prefix=/usr/local
-ENV PATH /opt/rh/devtoolset-10/root/usr/bin:$PATH
-
-# centos8 arm64 dependencies
-FROM --platform=linux/arm64 nvidia/cuda:11.4.3-devel-centos8 AS base-arm64
-RUN sed -i -e 's/mirrorlist/#mirrorlist/g' -e 's|#baseurl=http://mirror.centos.org|baseurl=http://vault.centos.org|g' /etc/yum.repos.d/CentOS-*
-RUN yum install -y git cmake
-
-FROM base-${TARGETARCH}
-ARG TARGETARCH
-
-# install go
-ADD https://dl.google.com/go/go1.21.1.linux-$TARGETARCH.tar.gz /tmp/go1.21.1.tar.gz
-RUN mkdir -p /usr/local && tar xz -C /usr/local </tmp/go1.21.1.tar.gz
-
-# build the final binary
-WORKDIR /go/src/github.com/jmorganca/ollama
-COPY . .
-ENV GOOS=linux
-ENV GOARCH=$TARGETARCH
-
-ARG VERSION=0.0.0
-ARG GOFLAGS="'-ldflags -w -s'"
-
-RUN /usr/local/go/bin/go generate ./... && \
-    /usr/local/go/bin/go build .

README.md

@@ -9,27 +9,19 @@
 
 [![Discord](https://dcbadge.vercel.app/api/server/ollama?style=flat&compact=true)](https://discord.gg/ollama)
 
-Get up and running with large language models locally.
+> Note: Ollama is in early preview. Please report any issues you find.
 
-### macOS
+Run, create, and share large language models (LLMs).
 
-[Download](https://ollama.ai/download/Ollama-darwin.zip) 
+## Download
 
-### Linux & WSL2
-
-```
-curl https://ollama.ai/install.sh | sh
-```
-
-[Manual install instructions](https://github.com/jmorganca/ollama/blob/main/docs/linux.md)
-
-### Windows 
-
-coming soon
+- [Download](https://ollama.ai/download) for macOS on Apple Silicon (Intel coming soon)
+- Download for Windows and Linux (coming soon)
+- Build [from source](#building)
 
 ## Quickstart
 
-To run and chat with [Llama 2](https://ollama.ai/library/llama2):
+To run and chat with [Llama 2](https://ai.meta.com/llama), the new model by Meta:
 
 ```
 ollama run llama2
@@ -37,50 +29,32 @@ ollama run llama2
 
 ## Model library
 
-Ollama supports a list of open-source models available on [ollama.ai/library](https://ollama.ai/library "ollama model library")
+`ollama` includes a library of open-source models:
 
-Here are some example open-source models that can be downloaded:
-
-| Model              | Parameters | Size  | Download                       |
-| ------------------ | ---------- | ----- | ------------------------------ |
-| Mistral            | 7B         | 4.1GB | `ollama run mistral`           |
-| Llama 2            | 7B         | 3.8GB | `ollama run llama2`            |
-| Code Llama         | 7B         | 3.8GB | `ollama run codellama`         |
-| Llama 2 Uncensored | 7B         | 3.8GB | `ollama run llama2-uncensored` |
-| Llama 2 13B        | 13B        | 7.3GB | `ollama run llama2:13b`        |
-| Llama 2 70B        | 70B        | 39GB  | `ollama run llama2:70b`        |
-| Orca Mini          | 3B         | 1.9GB | `ollama run orca-mini`         |
-| Vicuna             | 7B         | 3.8GB | `ollama run vicuna`            |
+| Model                    | Parameters | Size  | Download                    |
+| ------------------------ | ---------- | ----- | --------------------------- |
+| Llama2                   | 7B         | 3.8GB | `ollama pull llama2`        |
+| Llama2 13B               | 13B        | 7.3GB | `ollama pull llama2:13b`    |
+| Orca Mini                | 3B         | 1.9GB | `ollama pull orca`          |
+| Vicuna                   | 7B         | 3.8GB | `ollama pull vicuna`        |
+| Nous-Hermes              | 13B        | 7.3GB | `ollama pull nous-hermes`   |
+| Wizard Vicuna Uncensored | 13B        | 7.3GB | `ollama pull wizard-vicuna` |
 
 > Note: You should have at least 8 GB of RAM to run the 3B models, 16 GB to run the 7B models, and 32 GB to run the 13B models.
 
-## Customize your own model
+## Examples
 
-### Import from GGUF or GGML
+### Run a model
 
-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 run llama2
+>>> hi
+Hello! How can I help you today?
+```
 
-1. Create a file named Modelfile, and add a `FROM` instruction with the local filepath to the model you want to import.
+### Create a custom model
 
-   ```
-   FROM ./vicuna-33b.Q4_0.gguf
-   ```
-
-3. Create the model in Ollama
-
-   ```
-   ollama create name -f path_to_modelfile
-   ```
-
-5. Run the model
-
-   ```
-   ollama run name
-   ```
-
-### Customize a prompt
-
-Models from the Ollama library can be customized with a prompt. The example
+Pull a base model:
 
 ```
 ollama pull llama2
@@ -109,85 +83,44 @@ ollama run mario
 Hello! It's your friend Mario.
 ```
 
-For more examples, see the [examples](./examples) directory. For more information on working with a Modelfile, see the [Modelfile](./docs/modelfile.md) documentation.
+For more examples, see the [examples](./examples) directory.
 
-## CLI Reference
-
-### Create a model
-
-`ollama create` is used to create a model from a Modelfile.
-
-### Pull a model
+### Pull a model from the registry
 
 ```
-ollama pull llama2
+ollama pull orca
 ```
 
-> This command can also be used to update a local model. Only the diff will be pulled.
-
-### Remove a model
-
-```
-ollama rm llama2
-```
-
-### Copy a model
-
-```
-ollama cp llama2 my-llama2
-```
-
-### Multiline input
-
-For multiline input, you can wrap text with `"""`:
-
-```
->>> """Hello,
-... world!
-... """
-I'm a basic program that prints the famous "Hello, world!" message to the console.
-```
-
-### Pass in prompt as arguments
-
-```
-$ ollama run llama2 "summarize this file:" "$(cat README.md)"
- Ollama is a lightweight, extensible framework for building and running language models on the local machine. It provides a simple API for creating, running, and managing models, as well as a library of pre-built models that can be easily used in a variety of applications.
-```
-
-### List models on your computer
+### Listing local models
 
 ```
 ollama list
 ```
 
-### Start Ollama
+## Model packages
 
-`ollama serve` is used when you want to start ollama without running the desktop application.
+### Overview
+
+Ollama bundles model weights, configuration, and data into a single package, defined by a [Modelfile](./docs/modelfile.md).
+
+<picture>
+  <source media="(prefers-color-scheme: dark)" height="480" srcset="https://github.com/jmorganca/ollama/assets/251292/2fd96b5f-191b-45c1-9668-941cfad4eb70">
+  <img alt="logo" height="480" src="https://github.com/jmorganca/ollama/assets/251292/2fd96b5f-191b-45c1-9668-941cfad4eb70">
+</picture>
 
 ## Building
 
-Install `cmake` and `go`:
-
 ```
-brew install cmake
-brew install go
-```
-
-Then generate dependencies and build:
-
-```
-go generate ./...
 go build .
 ```
 
-Next, start the server:
+To run it start the server:
 
 ```
-./ollama serve
+./ollama serve &
 ```
 
-Finally, in a separate shell, run a model:
+Finally, run a model!
 
 ```
 ./ollama run llama2
@@ -195,29 +128,10 @@ Finally, in a separate shell, run a model:
 
 ## REST API
 
-> See the [API documentation](./docs/api.md) for all endpoints.
+### `POST /api/generate`
 
-Ollama has an API for running and managing models. For example to generate text from a model:
+Generate text from a model.
 
 ```
-curl -X POST http://localhost:11434/api/generate -d '{
-  "model": "llama2",
-  "prompt":"Why is the sky blue?"
-}'
+curl -X POST http://localhost:11434/api/generate -d '{"model": "llama2", "prompt":"Why is the sky blue?"}'
 ```
-
-## Community Integrations
-
-- [LangChain](https://python.langchain.com/docs/integrations/llms/ollama) and [LangChain.js](https://js.langchain.com/docs/modules/model_io/models/llms/integrations/ollama) with [example](https://js.langchain.com/docs/use_cases/question_answering/local_retrieval_qa)
-- [LlamaIndex](https://gpt-index.readthedocs.io/en/stable/examples/llm/ollama.html)
-- [Raycast extension](https://github.com/MassimilianoPasquini97/raycast_ollama)
-- [Discollama](https://github.com/mxyng/discollama) (Discord bot inside the Ollama discord channel)
-- [Continue](https://github.com/continuedev/continue)
-- [Obsidian Ollama plugin](https://github.com/hinterdupfinger/obsidian-ollama)
-- [Dagger Chatbot](https://github.com/samalba/dagger-chatbot)
-- [LiteLLM](https://github.com/BerriAI/litellm)
-- [Discord AI Bot](https://github.com/mekb-turtle/discord-ai-bot)
-- [HTML UI](https://github.com/rtcfirefly/ollama-ui)
-- [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)

api/client.go

@@ -9,27 +9,16 @@ import (
 	"io"
 	"net/http"
 	"net/url"
-	"os"
-	"runtime"
-	"strings"
-
-	"github.com/jmorganca/ollama/version"
-)
-
-const DefaultHost = "127.0.0.1:11434"
-
-var (
-	envHost = os.Getenv("OLLAMA_HOST")
 )
 
 type Client struct {
-	Base    url.URL
+	base    url.URL
 	HTTP    http.Client
 	Headers http.Header
 }
 
 func checkError(resp *http.Response, body []byte) error {
-	if resp.StatusCode < http.StatusBadRequest {
+	if resp.StatusCode >= 200 && resp.StatusCode < 400 {
 		return nil
 	}
 
@@ -44,34 +33,16 @@ func checkError(resp *http.Response, body []byte) error {
 	return apiError
 }
 
-// Host returns the default host to use for the client. It is determined in the following order:
-// 1. The OLLAMA_HOST environment variable
-// 2. The default host (localhost:11434)
-func Host() string {
-	if envHost != "" {
-		return envHost
-	}
-	return DefaultHost
-}
-
-// FromEnv creates a new client using Host() as the host. An error is returns
-// if the host is invalid.
-func FromEnv() (*Client, error) {
-	h := Host()
-	if !strings.HasPrefix(h, "http://") && !strings.HasPrefix(h, "https://") {
-		h = "http://" + h
+func NewClient(hosts ...string) *Client {
+	host := "127.0.0.1:11434"
+	if len(hosts) > 0 {
+		host = hosts[0]
 	}
 
-	u, err := url.Parse(h)
-	if err != nil {
-		return nil, fmt.Errorf("could not parse host: %w", err)
+	return &Client{
+		base: url.URL{Scheme: "http", Host: host},
+		HTTP: http.Client{},
 	}
-
-	if u.Port() == "" {
-		u.Host += ":11434"
-	}
-
-	return &Client{Base: *u, HTTP: http.Client{}}, nil
 }
 
 func (c *Client) do(ctx context.Context, method, path string, reqData, respData any) error {
@@ -86,21 +57,21 @@ func (c *Client) do(ctx context.Context, method, path string, reqData, respData
 		reqBody = bytes.NewReader(data)
 	}
 
-	requestURL := c.Base.JoinPath(path)
-	request, err := http.NewRequestWithContext(ctx, method, requestURL.String(), reqBody)
+	url := c.base.JoinPath(path).String()
+
+	req, err := http.NewRequestWithContext(ctx, method, url, reqBody)
 	if err != nil {
 		return err
 	}
 
-	request.Header.Set("Content-Type", "application/json")
-	request.Header.Set("Accept", "application/json")
-	request.Header.Set("User-Agent", fmt.Sprintf("ollama/%s (%s %s) Go/%s", version.Version, runtime.GOARCH, runtime.GOOS, runtime.Version()))
+	req.Header.Set("Content-Type", "application/json")
+	req.Header.Set("Accept", "application/json")
 
 	for k, v := range c.Headers {
-		request.Header[k] = v
+		req.Header[k] = v
 	}
 
-	respObj, err := c.HTTP.Do(request)
+	respObj, err := c.HTTP.Do(req)
 	if err != nil {
 		return err
 	}
@@ -134,15 +105,13 @@ func (c *Client) stream(ctx context.Context, method, path string, data any, fn f
 		buf = bytes.NewBuffer(bts)
 	}
 
-	requestURL := c.Base.JoinPath(path)
-	request, err := http.NewRequestWithContext(ctx, method, requestURL.String(), buf)
+	request, err := http.NewRequestWithContext(ctx, method, c.base.JoinPath(path).String(), buf)
 	if err != nil {
 		return err
 	}
 
 	request.Header.Set("Content-Type", "application/json")
 	request.Header.Set("Accept", "application/json")
-	request.Header.Set("User-Agent", fmt.Sprintf("ollama/%s (%s %s) Go/%s", version.Version, runtime.GOARCH, runtime.GOOS, runtime.Version()))
 
 	response, err := http.DefaultClient.Do(request)
 	if err != nil {
@@ -162,10 +131,10 @@ func (c *Client) stream(ctx context.Context, method, path string, data any, fn f
 		}
 
 		if errorResponse.Error != "" {
-			return fmt.Errorf(errorResponse.Error)
+			return fmt.Errorf("stream: %s", errorResponse.Error)
 		}
 
-		if response.StatusCode >= http.StatusBadRequest {
+		if response.StatusCode >= 400 {
 			return StatusError{
 				StatusCode:   response.StatusCode,
 				Status:       response.Status,
@@ -220,11 +189,11 @@ func (c *Client) Push(ctx context.Context, req *PushRequest, fn PushProgressFunc
 	})
 }
 
-type CreateProgressFunc func(ProgressResponse) error
+type CreateProgressFunc func(CreateProgress) error
 
 func (c *Client) Create(ctx context.Context, req *CreateRequest, fn CreateProgressFunc) error {
 	return c.stream(ctx, http.MethodPost, "/api/create", req, func(bts []byte) error {
-		var resp ProgressResponse
+		var resp CreateProgress
 		if err := json.Unmarshal(bts, &resp); err != nil {
 			return err
 		}
@@ -241,31 +210,9 @@ func (c *Client) List(ctx context.Context) (*ListResponse, error) {
 	return &lr, nil
 }
 
-func (c *Client) Copy(ctx context.Context, req *CopyRequest) error {
-	if err := c.do(ctx, http.MethodPost, "/api/copy", req, nil); err != nil {
-		return err
-	}
-	return nil
-}
-
 func (c *Client) Delete(ctx context.Context, req *DeleteRequest) error {
 	if err := c.do(ctx, http.MethodDelete, "/api/delete", req, nil); err != nil {
 		return err
 	}
 	return nil
 }
-
-func (c *Client) Show(ctx context.Context, req *ShowRequest) (*ShowResponse, error) {
-	var resp ShowResponse
-	if err := c.do(ctx, http.MethodPost, "/api/show", req, &resp); err != nil {
-		return nil, err
-	}
-	return &resp, nil
-}
-
-func (c *Client) Heartbeat(ctx context.Context) error {
-	if err := c.do(ctx, http.MethodHead, "/", nil, nil); err != nil {
-		return err
-	}
-	return nil
-}

api/client.py

@@ -1,225 +0,0 @@
-import os
-import json
-import requests
-
-BASE_URL = os.environ.get('OLLAMA_HOST', 'http://localhost:11434')
-
-# Generate a response for a given prompt with a provided model. This is a streaming endpoint, so will be a series of responses.
-# The final response object will include statistics and additional data from the request. Use the callback function to override
-# the default handler.
-def generate(model_name, prompt, system=None, template=None, context=None, options=None, callback=None):
-    try:
-        url = f"{BASE_URL}/api/generate"
-        payload = {
-            "model": model_name, 
-            "prompt": prompt, 
-            "system": system, 
-            "template": template, 
-            "context": context, 
-            "options": options
-        }
-        
-        # Remove keys with None values
-        payload = {k: v for k, v in payload.items() if v is not None}
-        
-        with requests.post(url, json=payload, stream=True) as response:
-            response.raise_for_status()
-            
-            # Creating a variable to hold the context history of the final chunk
-            final_context = None
-            
-            # Variable to hold concatenated response strings if no callback is provided
-            full_response = ""
-
-            # Iterating over the response line by line and displaying the details
-            for line in response.iter_lines():
-                if line:
-                    # Parsing each line (JSON chunk) and extracting the details
-                    chunk = json.loads(line)
-                    
-                    # If a callback function is provided, call it with the chunk
-                    if callback:
-                        callback(chunk)
-                    else:
-                        # If this is not the last chunk, add the "response" field value to full_response and print it
-                        if not chunk.get("done"):
-                            response_piece = chunk.get("response", "")
-                            full_response += response_piece
-                            print(response_piece, end="", flush=True)
-                    
-                    # Check if it's the last chunk (done is true)
-                    if chunk.get("done"):
-                        final_context = chunk.get("context")
-            
-            # Return the full response and the final context
-            return full_response, final_context
-    except requests.exceptions.RequestException as e:
-        print(f"An error occurred: {e}")
-        return None, None
-
-# Create a model from a Modelfile. Use the callback function to override the default handler.
-def create(model_name, model_path, callback=None):
-    try:
-        url = f"{BASE_URL}/api/create"
-        payload = {"name": model_name, "path": model_path}
-        
-        # Making a POST request with the stream parameter set to True to handle streaming responses
-        with requests.post(url, json=payload, stream=True) as response:
-            response.raise_for_status()
-
-            # Iterating over the response line by line and displaying the status
-            for line in response.iter_lines():
-                if line:
-                    # Parsing each line (JSON chunk) and extracting the status
-                    chunk = json.loads(line)
-
-                    if callback:
-                        callback(chunk)
-                    else:
-                        print(f"Status: {chunk.get('status')}")
-    except requests.exceptions.RequestException as e:
-        print(f"An error occurred: {e}")
-
-# Pull a model from a the model registry. Cancelled pulls are resumed from where they left off, and multiple
-# calls to will share the same download progress. Use the callback function to override the default handler.
-def pull(model_name, insecure=False, callback=None):
-    try:
-        url = f"{BASE_URL}/api/pull"
-        payload = {
-            "name": model_name,
-            "insecure": insecure
-        }
-
-        # Making a POST request with the stream parameter set to True to handle streaming responses
-        with requests.post(url, json=payload, stream=True) as response:
-            response.raise_for_status()
-
-            # Iterating over the response line by line and displaying the details
-            for line in response.iter_lines():
-                if line:
-                    # Parsing each line (JSON chunk) and extracting the details
-                    chunk = json.loads(line)
-
-                    # If a callback function is provided, call it with the chunk
-                    if callback:
-                        callback(chunk)
-                    else:
-                        # Print the status message directly to the console
-                        print(chunk.get('status', ''), end='', flush=True)
-                    
-                    # If there's layer data, you might also want to print that (adjust as necessary)
-                    if 'digest' in chunk:
-                        print(f" - Digest: {chunk['digest']}", end='', flush=True)
-                        print(f" - Total: {chunk['total']}", end='', flush=True)
-                        print(f" - Completed: {chunk['completed']}", end='\n', flush=True)
-                    else:
-                        print()
-    except requests.exceptions.RequestException as e:
-        print(f"An error occurred: {e}")
-
-# Push a model to the model registry. Use the callback function to override the default handler.
-def push(model_name, insecure=False, callback=None):
-    try:
-        url = f"{BASE_URL}/api/push"
-        payload = {
-            "name": model_name,
-            "insecure": insecure
-        }
-
-        # Making a POST request with the stream parameter set to True to handle streaming responses
-        with requests.post(url, json=payload, stream=True) as response:
-            response.raise_for_status()
-
-            # Iterating over the response line by line and displaying the details
-            for line in response.iter_lines():
-                if line:
-                    # Parsing each line (JSON chunk) and extracting the details
-                    chunk = json.loads(line)
-
-                    # If a callback function is provided, call it with the chunk
-                    if callback:
-                        callback(chunk)
-                    else:
-                        # Print the status message directly to the console
-                        print(chunk.get('status', ''), end='', flush=True)
-                    
-                    # If there's layer data, you might also want to print that (adjust as necessary)
-                    if 'digest' in chunk:
-                        print(f" - Digest: {chunk['digest']}", end='', flush=True)
-                        print(f" - Total: {chunk['total']}", end='', flush=True)
-                        print(f" - Completed: {chunk['completed']}", end='\n', flush=True)
-                    else:
-                        print()
-    except requests.exceptions.RequestException as e:
-        print(f"An error occurred: {e}")
-
-# List models that are available locally.
-def list():
-    try:
-        response = requests.get(f"{BASE_URL}/api/tags")
-        response.raise_for_status()
-        data = response.json()
-        models = data.get('models', [])
-        return models
-
-    except requests.exceptions.RequestException as e:
-        print(f"An error occurred: {e}")
-        return None
-
-# Copy a model. Creates a model with another name from an existing model.
-def copy(source, destination):
-    try:
-        # Create the JSON payload
-        payload = {
-            "source": source,
-            "destination": destination
-        }
-        
-        response = requests.post(f"{BASE_URL}/api/copy", json=payload)
-        response.raise_for_status()
-        
-        # If the request was successful, return a message indicating that the copy was successful
-        return "Copy successful"
-
-    except requests.exceptions.RequestException as e:
-        print(f"An error occurred: {e}")
-        return None
-
-# Delete a model and its data.
-def delete(model_name):
-    try:
-        url = f"{BASE_URL}/api/delete"
-        payload = {"name": model_name}
-        response = requests.delete(url, json=payload)
-        response.raise_for_status()
-        return "Delete successful"
-    except requests.exceptions.RequestException as e:
-        print(f"An error occurred: {e}")
-        return None
-
-# Show info about a model.
-def show(model_name):
-    try:
-        url = f"{BASE_URL}/api/show"
-        payload = {"name": model_name}
-        response = requests.post(url, json=payload)
-        response.raise_for_status()
-        
-        # Parse the JSON response and return it
-        data = response.json()
-        return data
-    except requests.exceptions.RequestException as e:
-        print(f"An error occurred: {e}")
-        return None
-
-def heartbeat():
-    try:
-        url = f"{BASE_URL}/"
-        response = requests.head(url)
-        response.raise_for_status()
-        return "Ollama is running"
-    except requests.exceptions.RequestException as e:
-        print(f"An error occurred: {e}")
-        return "Ollama is not running"
-
-

api/types.go

@@ -1,13 +1,9 @@
 package api
 
 import (
-	"encoding/json"
 	"fmt"
-	"log"
-	"math"
 	"os"
-	"reflect"
-	"strings"
+	"runtime"
 	"time"
 )
 
@@ -31,41 +27,12 @@ func (e StatusError) Error() string {
 	}
 }
 
-// /api/chat
-type Message struct {
-	Role    string `json:"role"`
-	Content string `json:"content"`
-}
-
-type ChatRequest struct {
-	Model    string    `json:"model"`
-	Messages []Message `json:"messages"`
-}
-
-type ChatResponse struct {
-	CreatedAt time.Time `json:"created_at"`
-	Message   Message   `json:"message"`
-}
-
 type GenerateRequest struct {
-	Model    string `json:"model"`
-	Prompt   string `json:"prompt"`
-	System   string `json:"system"`
-	Template string `json:"template"`
-	Context  []int  `json:"context,omitempty"`
+	Model   string `json:"model"`
+	Prompt  string `json:"prompt"`
+	Context []int  `json:"context,omitempty"`
 
-	Options map[string]interface{} `json:"options"`
-}
-
-type EmbeddingRequest struct {
-	Model  string `json:"model"`
-	Prompt string `json:"prompt"`
-
-	Options map[string]interface{} `json:"options"`
-}
-
-type EmbeddingResponse struct {
-	Embedding []float64 `json:"embedding"`
+	Options `json:"options"`
 }
 
 type CreateRequest struct {
@@ -73,57 +40,43 @@ type CreateRequest struct {
 	Path string `json:"path"`
 }
 
+type CreateProgress struct {
+	Status string `json:"status"`
+}
+
 type DeleteRequest struct {
 	Name string `json:"name"`
 }
 
-type ShowRequest struct {
-	Name string `json:"name"`
-}
-
-type ShowResponse struct {
-	License    string `json:"license,omitempty"`
-	Modelfile  string `json:"modelfile,omitempty"`
-	Parameters string `json:"parameters,omitempty"`
-	Template   string `json:"template,omitempty"`
-	System     string `json:"system,omitempty"`
-}
-
-type CopyRequest struct {
-	Source      string `json:"source"`
-	Destination string `json:"destination"`
-}
-
 type PullRequest struct {
 	Name     string `json:"name"`
 	Insecure bool   `json:"insecure,omitempty"`
+	Username string `json:"username"`
+	Password string `json:"password"`
 }
 
 type ProgressResponse struct {
 	Status    string `json:"status"`
 	Digest    string `json:"digest,omitempty"`
-	Total     int64  `json:"total,omitempty"`
-	Completed int64  `json:"completed,omitempty"`
+	Total     int    `json:"total,omitempty"`
+	Completed int    `json:"completed,omitempty"`
 }
 
 type PushRequest struct {
 	Name     string `json:"name"`
 	Insecure bool   `json:"insecure,omitempty"`
+	Username string `json:"username"`
+	Password string `json:"password"`
 }
 
 type ListResponse struct {
-	Models []ModelResponse `json:"models"`
+	Models []ListResponseModel `json:"models"`
 }
 
-type ModelResponse struct {
+type ListResponseModel struct {
 	Name       string    `json:"name"`
 	ModifiedAt time.Time `json:"modified_at"`
-	Size       int64     `json:"size"`
-	Digest     string    `json:"digest"`
-}
-
-type TokenResponse struct {
-	Token string `json:"token"`
+	Size       int       `json:"size"`
 }
 
 type GenerateResponse struct {
@@ -135,7 +88,6 @@ type GenerateResponse struct {
 	Context []int `json:"context,omitempty"`
 
 	TotalDuration      time.Duration `json:"total_duration,omitempty"`
-	LoadDuration       time.Duration `json:"load_duration,omitempty"`
 	PromptEvalCount    int           `json:"prompt_eval_count,omitempty"`
 	PromptEvalDuration time.Duration `json:"prompt_eval_duration,omitempty"`
 	EvalCount          int           `json:"eval_count,omitempty"`
@@ -147,10 +99,6 @@ func (r *GenerateResponse) Summary() {
 		fmt.Fprintf(os.Stderr, "total duration:       %v\n", r.TotalDuration)
 	}
 
-	if r.LoadDuration > 0 {
-		fmt.Fprintf(os.Stderr, "load duration:        %v\n", r.LoadDuration)
-	}
-
 	if r.PromptEvalCount > 0 {
 		fmt.Fprintf(os.Stderr, "prompt eval count:    %d token(s)\n", r.PromptEvalCount)
 	}
@@ -177,143 +125,50 @@ type Options struct {
 	UseNUMA bool `json:"numa,omitempty"`
 
 	// Model options
-	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"`
-	MainGPU            int     `json:"main_gpu,omitempty"`
-	LowVRAM            bool    `json:"low_vram,omitempty"`
-	F16KV              bool    `json:"f16_kv,omitempty"`
-	LogitsAll          bool    `json:"logits_all,omitempty"`
-	VocabOnly          bool    `json:"vocab_only,omitempty"`
-	UseMMap            bool    `json:"use_mmap,omitempty"`
-	UseMLock           bool    `json:"use_mlock,omitempty"`
-	EmbeddingOnly      bool    `json:"embedding_only,omitempty"`
-	RopeFrequencyBase  float32 `json:"rope_frequency_base,omitempty"`
-	RopeFrequencyScale float32 `json:"rope_frequency_scale,omitempty"`
+	NumCtx        int  `json:"num_ctx,omitempty"`
+	NumBatch      int  `json:"num_batch,omitempty"`
+	NumGPU        int  `json:"num_gpu,omitempty"`
+	MainGPU       int  `json:"main_gpu,omitempty"`
+	LowVRAM       bool `json:"low_vram,omitempty"`
+	F16KV         bool `json:"f16_kv,omitempty"`
+	LogitsAll     bool `json:"logits_all,omitempty"`
+	VocabOnly     bool `json:"vocab_only,omitempty"`
+	UseMMap       bool `json:"use_mmap,omitempty"`
+	UseMLock      bool `json:"use_mlock,omitempty"`
+	EmbeddingOnly bool `json:"embedding_only,omitempty"`
 
 	// Predict options
-	NumPredict       int      `json:"num_predict,omitempty"`
-	TopK             int      `json:"top_k,omitempty"`
-	TopP             float32  `json:"top_p,omitempty"`
-	TFSZ             float32  `json:"tfs_z,omitempty"`
-	TypicalP         float32  `json:"typical_p,omitempty"`
-	RepeatLastN      int      `json:"repeat_last_n,omitempty"`
-	Temperature      float32  `json:"temperature,omitempty"`
-	RepeatPenalty    float32  `json:"repeat_penalty,omitempty"`
-	PresencePenalty  float32  `json:"presence_penalty,omitempty"`
-	FrequencyPenalty float32  `json:"frequency_penalty,omitempty"`
-	Mirostat         int      `json:"mirostat,omitempty"`
-	MirostatTau      float32  `json:"mirostat_tau,omitempty"`
-	MirostatEta      float32  `json:"mirostat_eta,omitempty"`
-	PenalizeNewline  bool     `json:"penalize_newline,omitempty"`
-	Stop             []string `json:"stop,omitempty"`
+	RepeatLastN      int     `json:"repeat_last_n,omitempty"`
+	RepeatPenalty    float32 `json:"repeat_penalty,omitempty"`
+	FrequencyPenalty float32 `json:"frequency_penalty,omitempty"`
+	PresencePenalty  float32 `json:"presence_penalty,omitempty"`
+	Temperature      float32 `json:"temperature,omitempty"`
+	TopK             int     `json:"top_k,omitempty"`
+	TopP             float32 `json:"top_p,omitempty"`
+	TFSZ             float32 `json:"tfs_z,omitempty"`
+	TypicalP         float32 `json:"typical_p,omitempty"`
+	Mirostat         int     `json:"mirostat,omitempty"`
+	MirostatTau      float32 `json:"mirostat_tau,omitempty"`
+	MirostatEta      float32 `json:"mirostat_eta,omitempty"`
 
 	NumThread int `json:"num_thread,omitempty"`
 }
 
-func (opts *Options) FromMap(m map[string]interface{}) error {
-	valueOpts := reflect.ValueOf(opts).Elem() // names of the fields in the options struct
-	typeOpts := reflect.TypeOf(opts).Elem()   // types of the fields in the options struct
-
-	// build map of json struct tags to their types
-	jsonOpts := make(map[string]reflect.StructField)
-	for _, field := range reflect.VisibleFields(typeOpts) {
-		jsonTag := strings.Split(field.Tag.Get("json"), ",")[0]
-		if jsonTag != "" {
-			jsonOpts[jsonTag] = field
-		}
-	}
-
-	for key, val := range m {
-		if opt, ok := jsonOpts[key]; ok {
-			field := valueOpts.FieldByName(opt.Name)
-			if field.IsValid() && field.CanSet() {
-				if val == nil {
-					continue
-				}
-
-				switch field.Kind() {
-				case reflect.Int:
-					switch t := val.(type) {
-					case int64:
-						field.SetInt(t)
-					case float64:
-						// when JSON unmarshals numbers, it uses float64, not int
-						field.SetInt(int64(t))
-					default:
-						log.Printf("could not convert model parameter %v to int, skipped", key)
-					}
-				case reflect.Bool:
-					val, ok := val.(bool)
-					if !ok {
-						log.Printf("could not convert model parameter %v to bool, skipped", key)
-						continue
-					}
-					field.SetBool(val)
-				case reflect.Float32:
-					// JSON unmarshals to float64
-					val, ok := val.(float64)
-					if !ok {
-						log.Printf("could not convert model parameter %v to float32, skipped", key)
-						continue
-					}
-					field.SetFloat(val)
-				case reflect.String:
-					val, ok := val.(string)
-					if !ok {
-						log.Printf("could not convert model parameter %v to string, skipped", key)
-						continue
-					}
-					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 to slice, skipped", key)
-						continue
-					}
-					// 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 to slice of strings, skipped", key)
-							continue
-						}
-						slice[i] = str
-					}
-					field.Set(reflect.ValueOf(slice))
-				default:
-					return fmt.Errorf("unknown type loading config params: %v", field.Kind())
-				}
-			}
-		}
-	}
-	return nil
-}
-
 func DefaultOptions() Options {
 	return Options{
 		Seed: -1,
 
 		UseNUMA: false,
 
-		NumCtx:             2048,
-		NumKeep:            -1,
-		NumBatch:           512,
-		NumGPU:             -1, // -1 here indicates that NumGPU should be set dynamically
-		NumGQA:             1,
-		LowVRAM:            false,
-		F16KV:              true,
-		UseMMap:            true,
-		UseMLock:           false,
-		RopeFrequencyBase:  10000.0,
-		RopeFrequencyScale: 1.0,
-		EmbeddingOnly:      true,
+		NumCtx:   2048,
+		NumBatch: 512,
+		NumGPU:   1,
+		LowVRAM:  false,
+		F16KV:    true,
+		UseMMap:  true,
+		UseMLock: false,
 
-		RepeatLastN:      64,
+		RepeatLastN:      512,
 		RepeatPenalty:    1.1,
 		FrequencyPenalty: 0.0,
 		PresencePenalty:  0.0,
@@ -325,37 +180,7 @@ func DefaultOptions() Options {
 		Mirostat:         0,
 		MirostatTau:      5.0,
 		MirostatEta:      0.1,
-		PenalizeNewline:  true,
 
-		NumThread: 0, // let the runtime decide
+		NumThread: runtime.NumCPU(),
 	}
 }
-
-type Duration struct {
-	time.Duration
-}
-
-func (d *Duration) UnmarshalJSON(b []byte) (err error) {
-	var v any
-	if err := json.Unmarshal(b, &v); err != nil {
-		return err
-	}
-
-	d.Duration = 5 * time.Minute
-
-	switch t := v.(type) {
-	case float64:
-		if t < 0 {
-			t = math.MaxFloat64
-		}
-
-		d.Duration = time.Duration(t)
-	case string:
-		d.Duration, err = time.ParseDuration(t)
-		if err != nil {
-			return err
-		}
-	}
-
-	return nil
-}

app/README.md

@@ -1,5 +1,7 @@
 # Desktop
 
+_Note: the Ollama desktop app is a work in progress and is not ready yet for general use._
+
 This app builds upon Ollama to provide a desktop experience for running models.
 
 ## Developing
@@ -7,15 +9,19 @@ This app builds upon Ollama to provide a desktop experience for running models.
 First, build the `ollama` binary:
 
 ```
-cd ..
-go build .
+make -C ..
 ```
 
 Then run the desktop app with `npm start`:
 
 ```
-cd app
 npm install
 npm start
 ```
 
+## Coming soon
+
+- Browse the latest available models on Hugging Face and other sources
+- Keep track of previous conversations with models
+- Switch quickly between models
+- Connect to remote Ollama servers to run models

app/forge.config.ts

@@ -18,15 +18,12 @@ const config: ForgeConfig = {
     asar: true,
     icon: './assets/icon.icns',
     extraResource: [
-      '../dist/ollama',
-      path.join(__dirname, './assets/iconTemplate.png'),
-      path.join(__dirname, './assets/iconTemplate@2x.png'),
-      path.join(__dirname, './assets/iconUpdateTemplate.png'),
-      path.join(__dirname, './assets/iconUpdateTemplate@2x.png'),
-      path.join(__dirname, './assets/iconDarkTemplate.png'),
-      path.join(__dirname, './assets/iconDarkTemplate@2x.png'),
-      path.join(__dirname, './assets/iconDarkUpdateTemplate.png'),
-      path.join(__dirname, './assets/iconDarkUpdateTemplate@2x.png'),
+      '../ollama',
+      path.join(__dirname, './assets/ollama_icon_16x16Template.png'),
+      path.join(__dirname, './assets/ollama_icon_16x16Template@2x.png'),
+      path.join(__dirname, './assets/ollama_outline_icon_16x16Template.png'),
+      path.join(__dirname, './assets/ollama_outline_icon_16x16Template@2x.png'),
+      ...(process.platform === 'darwin' ? ['../llama/ggml-metal.metal'] : []),
     ],
     ...(process.env.SIGN
       ? {
@@ -41,9 +38,6 @@ const config: ForgeConfig = {
           },
         }
       : {}),
-    osxUniversal: {
-      x64ArchFiles: '**/ollama',
-    },
   },
   rebuildConfig: {},
   makers: [new MakerSquirrel({}), new MakerZIP({}, ['darwin'])],

app/package-lock.json

@@ -32,7 +32,6 @@
         "@electron-forge/plugin-auto-unpack-natives": "^6.2.1",
         "@electron-forge/plugin-webpack": "^6.2.1",
         "@electron-forge/publisher-github": "^6.2.1",
-        "@electron/universal": "^1.4.1",
         "@svgr/webpack": "^8.0.1",
         "@types/chmodr": "^1.0.0",
         "@types/node": "^20.4.0",
@@ -3329,9 +3328,9 @@
       }
     },
     "node_modules/@electron/universal": {
-      "version": "1.4.1",
-      "resolved": "https://registry.npmjs.org/@electron/universal/-/universal-1.4.1.tgz",
-      "integrity": "sha512-lE/U3UNw1YHuowNbTmKNs9UlS3En3cPgwM5MI+agIgr/B1hSze9NdOP0qn7boZaI9Lph8IDv3/24g9IxnJP7aQ==",
+      "version": "1.3.4",
+      "resolved": "https://registry.npmjs.org/@electron/universal/-/universal-1.3.4.tgz",
+      "integrity": "sha512-BdhBgm2ZBnYyYRLRgOjM5VHkyFItsbggJ0MHycOjKWdFGYwK97ZFXH54dTvUWEfha81vfvwr5On6XBjt99uDcg==",
       "dev": true,
       "dependencies": {
         "@electron/asar": "^3.2.1",

app/package.json

@@ -6,14 +6,12 @@
   "main": ".webpack/main",
   "scripts": {
     "start": "electron-forge start",
-    "package": "electron-forge package --arch universal",
-    "package:sign": "SIGN=1 electron-forge package --arch universal",
-    "make": "electron-forge make --arch universal",
-    "make:sign": "SIGN=1 electron-forge make --arch universal",
+    "package": "electron-forge package",
+    "package:sign": "SIGN=1 electron-forge package",
+    "make": "electron-forge make",
+    "make:sign": "SIGN=1 electron-forge make",
     "publish": "SIGN=1 electron-forge publish",
-    "lint": "eslint --ext .ts,.tsx .",
-    "format": "prettier --check . --ignore-path .gitignore",
-    "format:fix": "prettier --write . --ignore-path .gitignore"
+    "lint": "eslint --ext .ts,.tsx ."
   },
   "keywords": [],
   "author": {
@@ -32,7 +30,6 @@
     "@electron-forge/plugin-auto-unpack-natives": "^6.2.1",
     "@electron-forge/plugin-webpack": "^6.2.1",
     "@electron-forge/publisher-github": "^6.2.1",
-    "@electron/universal": "^1.4.1",
     "@svgr/webpack": "^8.0.1",
     "@types/chmodr": "^1.0.0",
     "@types/node": "^20.4.0",

app/src/app.tsx

@@ -2,7 +2,7 @@ import { useState } from 'react'
 import copy from 'copy-to-clipboard'
 import { CheckIcon, DocumentDuplicateIcon } from '@heroicons/react/24/outline'
 import Store from 'electron-store'
-import { getCurrentWindow, app } from '@electron/remote'
+import { getCurrentWindow } from '@electron/remote'
 
 import { install } from './install'
 import OllamaIcon from './ollama.svg'
@@ -51,15 +51,10 @@ export default function () {
               <div className='mx-auto'>
                 <button
                   onClick={async () => {
-                    try {
-                      await install()
-                      setStep(Step.FINISH)
-                    } catch (e) {
-                      console.error('could not install: ', e)
-                    } finally {
-                      getCurrentWindow().show()
-                      getCurrentWindow().focus()
-                    }
+                    await install()
+                    getCurrentWindow().show()
+                    getCurrentWindow().focus()
+                    setStep(Step.FINISH)
                   }}
                   className='no-drag rounded-dm mx-auto w-[60%] rounded-md bg-black px-4 py-2 text-sm text-white hover:brightness-110'
                 >

app/src/declarations.d.ts

@@ -1,4 +1,4 @@
 declare module '*.svg' {
-  const content: string
-  export default content
-}
+  const content: string;
+  export default content;
+}

app/src/index.ts

@@ -1,21 +1,17 @@
-import { spawn, ChildProcess } from 'child_process'
-import { app, autoUpdater, dialog, Tray, Menu, BrowserWindow, MenuItemConstructorOptions, nativeTheme } from 'electron'
+import { spawn } from 'child_process'
+import { app, autoUpdater, dialog, Tray, Menu, BrowserWindow, nativeTheme } from 'electron'
 import Store from 'electron-store'
 import winston from 'winston'
 import 'winston-daily-rotate-file'
 import * as path from 'path'
 
-import { v4 as uuidv4 } from 'uuid'
+import { analytics, id } from './telemetry'
 import { installed } from './install'
 
 require('@electron/remote/main').initialize()
 
-if (require('electron-squirrel-startup')) {
-  app.quit()
-}
-
 const store = new Store()
-
+let tray: Tray | null = null
 let welcomeWindow: BrowserWindow | null = null
 
 declare const MAIN_WINDOW_WEBPACK_ENTRY: string
@@ -32,30 +28,10 @@ const logger = winston.createLogger({
   format: winston.format.printf(info => info.message),
 })
 
-app.on('ready', () => {
-  const gotTheLock = app.requestSingleInstanceLock()
-  if (!gotTheLock) {
-    app.exit(0)
-    return
-  }
-
-  app.on('second-instance', () => {
-    if (app.hasSingleInstanceLock()) {
-      app.releaseSingleInstanceLock()
-    }
-
-    if (proc) {
-      proc.off('exit', restart)
-      proc.kill()
-    }
-
-    app.exit(0)
-  })
-
-  app.focus({ steal: true })
-
-  init()
-})
+const SingleInstanceLock = app.requestSingleInstanceLock()
+if (!SingleInstanceLock) {
+  app.quit()
+}
 
 function firstRunWindow() {
   // Create the browser window.
@@ -71,74 +47,65 @@ function firstRunWindow() {
       nodeIntegration: true,
       contextIsolation: false,
     },
+    alwaysOnTop: true,
   })
 
   require('@electron/remote/main').enable(welcomeWindow.webContents)
 
+  // and load the index.html of the app.
   welcomeWindow.loadURL(MAIN_WINDOW_WEBPACK_ENTRY)
+
   welcomeWindow.on('ready-to-show', () => welcomeWindow.show())
-  welcomeWindow.on('closed', () => {
-    if (process.platform === 'darwin') {
-      app.dock.hide()
+
+  // for debugging
+  // welcomeWindow.webContents.openDevTools()
+
+  if (process.platform === 'darwin') {
+    app.dock.hide()
+  }
+}
+
+function createSystemtray() {
+  let iconPath = nativeTheme.shouldUseDarkColors
+    ? path.join(__dirname, '..', '..', 'assets', 'ollama_icon_16x16Template.png') 
+    : path.join(__dirname, '..', '..', 'assets', 'ollama_outline_icon_16x16Template.png')
+
+  if (app.isPackaged) {
+    iconPath = nativeTheme.shouldUseDarkColors
+    ? path.join(process.resourcesPath, 'ollama_icon_16x16Template.png') 
+    : path.join(process.resourcesPath, 'ollama_outline_icon_16x16Template.png')
+  }
+
+  tray = new Tray(iconPath)
+
+  nativeTheme.on('updated', function theThemeHasChanged () {
+    if (nativeTheme.shouldUseDarkColors) {
+      app.isPackaged 
+        ? tray.setImage(path.join(process.resourcesPath, 'ollama_icon_16x16Template.png')) 
+        : tray.setImage(path.join(__dirname, '..', '..', 'assets', 'ollama_icon_16x16Template.png'))
+    } else {
+      app.isPackaged 
+        ? tray.setImage(path.join(process.resourcesPath, 'ollama_outline_icon_16x16Template.png')) 
+        : tray.setImage(path.join(__dirname, '..', '..', 'assets', 'ollama_outline_icon_16x16Template.png'))
     }
   })
+
+  const contextMenu = Menu.buildFromTemplate([{ role: 'quit', label: 'Quit Ollama', accelerator: 'Command+Q' }])
+
+  tray.setContextMenu(contextMenu)
+  tray.setToolTip('Ollama')
 }
 
-let tray: Tray | null = null
-let updateAvailable = false
-const assetPath = app.isPackaged ? process.resourcesPath : path.join(__dirname, '..', '..', 'assets')
-
-function trayIconPath() {
-  return nativeTheme.shouldUseDarkColors
-    ? updateAvailable
-      ? path.join(assetPath, 'iconDarkUpdateTemplate.png')
-      : path.join(assetPath, 'iconDarkTemplate.png')
-    : updateAvailable
-    ? path.join(assetPath, 'iconUpdateTemplate.png')
-    : path.join(assetPath, 'iconTemplate.png')
+if (require('electron-squirrel-startup')) {
+  app.quit()
 }
 
-function updateTrayIcon() {
-  if (tray) {
-    tray.setImage(trayIconPath())
-  }
-}
-
-function updateTray() {
-  const updateItems: MenuItemConstructorOptions[] = [
-    { label: 'An update is available', enabled: false },
-    {
-      label: 'Restart to update',
-      click: () => autoUpdater.quitAndInstall(),
-    },
-    { type: 'separator' },
-  ]
-
-  const menu = Menu.buildFromTemplate([
-    ...(updateAvailable ? updateItems : []),
-    { role: 'quit', label: 'Quit Ollama', accelerator: 'Command+Q' },
-  ])
-
-  if (!tray) {
-    tray = new Tray(trayIconPath())
-  }
-
-  tray.setToolTip(updateAvailable ? 'An update is available' : 'Ollama')
-  tray.setContextMenu(menu)
-  tray.setImage(trayIconPath())
-
-  nativeTheme.off('updated', updateTrayIcon)
-  nativeTheme.on('updated', updateTrayIcon)
-}
-
-let proc: ChildProcess = null
-
 function server() {
   const binary = app.isPackaged
     ? path.join(process.resourcesPath, 'ollama')
     : path.resolve(process.cwd(), '..', 'ollama')
 
-  proc = spawn(binary, ['serve'])
+  const proc = spawn(binary, ['serve'])
 
   proc.stdout.on('data', data => {
     logger.info(data.toString().trim())
@@ -148,32 +115,24 @@ function server() {
     logger.error(data.toString().trim())
   })
 
+  function restart() {
+    logger.info('Restarting the server...')
+    server()
+  }
+
   proc.on('exit', restart)
-}
 
-function restart() {
-  setTimeout(server, 1000)
-}
-
-app.on('before-quit', () => {
-  if (proc) {
+  app.on('before-quit', () => {
     proc.off('exit', restart)
-    proc.kill('SIGINT') // send SIGINT signal to the server, which also stops any loaded llms
-  }
-})
+    proc.kill()
+  })
+}
 
-function init() {
-  if (app.isPackaged) {
-    autoUpdater.checkForUpdates()
-    setInterval(() => {
-      if (!updateAvailable) {
-        autoUpdater.checkForUpdates()
-      }
-    }, 60 * 60 * 1000)
-  }
-
-  updateTray()
+if (process.platform === 'darwin') {
+  app.dock.hide()
+}
 
+app.on('ready', () => {
   if (process.platform === 'darwin') {
     if (app.isPackaged) {
       if (!app.isInApplicationsFolder()) {
@@ -209,13 +168,10 @@ function init() {
     }
   }
 
+  createSystemtray()
   server()
 
   if (store.get('first-time-run') && installed()) {
-    if (process.platform === 'darwin') {
-      app.dock.hide()
-    }
-
     app.setLoginItemSettings({ openAtLogin: app.getLoginItemSettings().openAtLogin })
     return
   }
@@ -223,7 +179,7 @@ function init() {
   // This is the first run or the CLI is no longer installed
   app.setLoginItemSettings({ openAtLogin: true })
   firstRunWindow()
-}
+})
 
 // Quit when all windows are closed, except on macOS. There, it's common
 // for applications and their menu bar to stay active until the user quits
@@ -234,30 +190,45 @@ app.on('window-all-closed', () => {
   }
 })
 
-function id(): string {
-  const id = store.get('id') as string
+// In this file you can include the rest of your app's specific main process
+// code. You can also put them in separate files and import them here.
+autoUpdater.setFeedURL({
+  url: `https://ollama.ai/api/update?os=${process.platform}&arch=${process.arch}&version=${app.getVersion()}`,
+})
 
-  if (id) {
-    return id
-  }
-
-  const uuid = uuidv4()
-  store.set('id', uuid)
-  return uuid
+async function heartbeat() {
+  analytics.track({
+    anonymousId: id(),
+    event: 'heartbeat',
+    properties: {
+      version: app.getVersion(),
+    },
+  })
 }
 
-autoUpdater.setFeedURL({
-  url: `https://ollama.ai/api/update?os=${process.platform}&arch=${
-    process.arch
-  }&version=${app.getVersion()}&id=${id()}`,
-})
+if (app.isPackaged) {
+  heartbeat()
+  autoUpdater.checkForUpdates()
+  setInterval(() => {
+    heartbeat()
+    autoUpdater.checkForUpdates()
+  }, 60 * 60 * 1000)
+}
 
 autoUpdater.on('error', e => {
   logger.error(`update check failed - ${e.message}`)
-  console.error(`update check failed - ${e.message}`)
 })
 
-autoUpdater.on('update-downloaded', () => {
-  updateAvailable = true
-  updateTray()
+autoUpdater.on('update-downloaded', (event, releaseNotes, releaseName) => {
+  dialog
+    .showMessageBox({
+      type: 'info',
+      buttons: ['Restart Now', 'Later'],
+      title: 'New update available',
+      message: process.platform === 'win32' ? releaseNotes : releaseName,
+      detail: 'A new version of Ollama is available. Restart to apply the update.',
+    })
+    .then(returnValue => {
+      if (returnValue.response === 0) autoUpdater.quitAndInstall()
+    })
 })

app/src/install.ts

@@ -15,7 +15,12 @@ export function installed() {
 export async function install() {
   const command = `do shell script "mkdir -p ${path.dirname(
     symlinkPath
-  )} && ln -F -s \\"${ollama}\\" \\"${symlinkPath}\\"" with administrator privileges`
+  )} && ln -F -s ${ollama} ${symlinkPath}" with administrator privileges`
 
-  await exec(`osascript -e '${command}'`)
+  try {
+    await exec(`osascript -e '${command}'`)
+  } catch (error) {
+    console.error(`cli: failed to install cli: ${error.message}`)
+    return
+  }
 }

app/src/telemetry.ts

@@ -0,0 +1,19 @@
+import { Analytics } from '@segment/analytics-node'
+import { v4 as uuidv4 } from 'uuid'
+import Store from 'electron-store'
+
+const store = new Store()
+
+export const analytics = new Analytics({ writeKey: process.env.TELEMETRY_WRITE_KEY || '<empty>' })
+
+export function id(): string {
+  const id = store.get('id') as string
+
+  if (id) {
+    return id
+  }
+
+  const uuid = uuidv4()
+  store.set('id', uuid)
+  return uuid
+}

docs/README.md

@@ -1,6 +0,0 @@
-# Documentation
-
-- [Modelfile](./modelfile.md)
-- [How to develop Ollama](./development.md)
-- [API](./api.md)
-- [Tutorials](./tutorials.md)

docs/api.md

@@ -1,355 +0,0 @@
-# API
-
-## Endpoints
-
-- [Generate a completion](#generate-a-completion)
-- [Create a Model](#create-a-model)
-- [List Local Models](#list-local-models)
-- [Show Model Information](#show-model-information)
-- [Copy a Model](#copy-a-model)
-- [Delete a Model](#delete-a-model)
-- [Pull a Model](#pull-a-model)
-- [Push a Model](#push-a-model)
-- [Generate Embeddings](#generate-embeddings)
-
-
-## Conventions
-
-### Model names
-
-Model names follow a `model:tag` format. Some examples are `orca-mini:3b-q4_1` and `llama2:70b`. The tag is optional and, if not provided, will default to `latest`. The tag is used to identify a specific version.
-
-### Durations
-
-All durations are returned in nanoseconds.
-
-### Streaming responses
-
-Certain endpoints stream responses as JSON objects delineated with the newline (`\n`) character.
-
-## Generate a completion
-
-```shell
-POST /api/generate
-```
-
-Generate a response for a given prompt with a provided model. This is a streaming endpoint, so will be a series of responses. The final response object will include statistics and additional data from the request.
-
-### Parameters
-
-- `model`: (required) the [model name](#model-names)
-- `prompt`: the prompt to generate a response for
-
-Advanced parameters:
-
-- `options`: additional model parameters listed in the documentation for the [Modelfile](./modelfile.md#valid-parameters-and-values) such as `temperature`
-- `system`: system prompt to (overrides what is defined in the `Modelfile`)
-- `template`: the full prompt or prompt template (overrides what is defined in the `Modelfile`)
-- `context`: the context parameter returned from a previous request to `/generate`, this can be used to keep a short conversational memory
-
-### Request
-
-```shell
-curl -X POST http://localhost:11434/api/generate -d '{
-  "model": "llama2:7b",
-  "prompt": "Why is the sky blue?"
-}'
-```
-
-### Response
-
-A stream of JSON objects:
-
-```json
-{
-  "model": "llama2:7b",
-  "created_at": "2023-08-04T08:52:19.385406455-07:00",
-  "response": "The",
-  "done": false
-}
-```
-
-The final response in the stream also includes additional data about the generation:
-
-- `total_duration`: time spent generating the response
-- `load_duration`: time spent in nanoseconds loading the model
-- `sample_count`: number of samples generated
-- `sample_duration`: time spent generating samples
-- `prompt_eval_count`: number of tokens in the prompt
-- `prompt_eval_duration`: time spent in nanoseconds evaluating the prompt
-- `eval_count`: number of tokens the response
-- `eval_duration`: time in nanoseconds spent generating the response
-- `context`: an encoding of the conversation used in this response, this can be sent in the next request to keep a conversational memory
-
-To calculate how fast the response is generated in tokens per second (token/s), divide `eval_count` / `eval_duration`.
-
-```json
-{
-  "model": "llama2:7b",
-  "created_at": "2023-08-04T19:22:45.499127Z",
-  "context": [1, 2, 3],
-  "done": true,
-  "total_duration": 5589157167,
-  "load_duration": 3013701500,
-  "sample_count": 114,
-  "sample_duration": 81442000,
-  "prompt_eval_count": 46,
-  "prompt_eval_duration": 1160282000,
-  "eval_count": 113,
-  "eval_duration": 1325948000
-}
-```
-
-## Create a Model
-
-```shell
-POST /api/create
-```
-
-Create a model from a [`Modelfile`](./modelfile.md)
-
-### Parameters
-
-- `name`: name of the model to create
-- `path`: path to the Modelfile
-
-### Request
-
-```shell
-curl -X POST http://localhost:11434/api/create -d '{
-  "name": "mario",
-  "path": "~/Modelfile"
-}'
-```
-
-### Response
-
-A stream of JSON objects. When finished, `status` is `success`.
-
-```json
-{
-  "status": "parsing modelfile"
-}
-```
-
-## List Local Models
-
-```shell
-GET /api/tags
-```
-
-List models that are available locally.
-
-### Request
-
-```shell
-curl http://localhost:11434/api/tags
-```
-
-### Response
-
-```json
-{
-  "models": [
-    {
-      "name": "llama2:7b",
-      "modified_at": "2023-08-02T17:02:23.713454393-07:00",
-      "size": 3791730596
-    },
-    {
-      "name": "llama2:13b",
-      "modified_at": "2023-08-08T12:08:38.093596297-07:00",
-      "size": 7323310500
-    }
-  ]
-}
-```
-
-## Show Model Information
-
-```shell
-POST /api/show
-```
-
-Show details about a model including modelfile, template, parameters, license, and system prompt.
-
-### Parameters
-
-- `name`: name of the model to show
-
-### Request
-
-```shell  
-curl http://localhost:11434/api/show -d '{
-  "name": "llama2:7b"
-}'
-```
-
-### Response
-
-```json
-{
-    "license": "<contents of license block>",
-    "modelfile": "# Modelfile generated by \"ollama show\"\n# To build a new Modelfile based on this one, replace the FROM line with:\n# FROM llama2:latest\n\nFROM /Users/username/.ollama/models/blobs/sha256:8daa9615cce30c259a9555b1cc250d461d1bc69980a274b44d7eda0be78076d8\nTEMPLATE \"\"\"[INST] {{ if and .First .System }}<<SYS>>{{ .System }}<</SYS>>\n\n{{ end }}{{ .Prompt }} [/INST] \"\"\"\nSYSTEM \"\"\"\"\"\"\nPARAMETER stop [INST]\nPARAMETER stop [/INST]\nPARAMETER stop <<SYS>>\nPARAMETER stop <</SYS>>\n",
-    "parameters": "stop                           [INST]\nstop                           [/INST]\nstop                           <<SYS>>\nstop                           <</SYS>>",
-    "template": "[INST] {{ if and .First .System }}<<SYS>>{{ .System }}<</SYS>>\n\n{{ end }}{{ .Prompt }} [/INST] "
-}
-```
-
-## Copy a Model
-
-```shell
-POST /api/copy
-```
-
-Copy a model. Creates a model with another name from an existing model.
-
-### Request
-
-```shell
-curl http://localhost:11434/api/copy -d '{
-  "source": "llama2:7b",
-  "destination": "llama2-backup"
-}'
-```
-
-## Delete a Model
-
-```shell
-DELETE /api/delete
-```
-
-Delete a model and its data.
-
-### Parameters
-
-- `model`: model name to delete
-
-### Request
-
-```shell
-curl -X DELETE http://localhost:11434/api/delete -d '{
-  "name": "llama2:13b"
-}'
-```
-
-## Pull a Model
-
-```shell
-POST /api/pull
-```
-
-Download a model from the ollama library. Cancelled pulls are resumed from where they left off, and multiple calls will share the same download progress.
-
-### Parameters
-
-- `name`: name of the model to pull
-- `insecure`: (optional) allow insecure connections to the library. Only use this if you are pulling from your own library during development.
-
-### Request
-
-```shell
-curl -X POST http://localhost:11434/api/pull -d '{
-  "name": "llama2:7b"
-}'
-```
-
-### Response
-
-```json
-{
-  "status": "downloading digestname",
-  "digest": "digestname",
-  "total": 2142590208
-}
-```
-
-## Push a Model
-
-```shell
-POST /api/push
-```
-
-Upload a model to a model library. Requires registering for ollama.ai and adding a public key first.
-
-### Parameters
-
-- `name`: name of the model to push in the form of `<namespace>/<model>:<tag>`
-- `insecure`: (optional) allow insecure connections to the library. Only use this if you are pushing to your library during development.  
-
-### Request
-
-```shell
-curl -X POST http://localhost:11434/api/push -d '{
-  "name": "mattw/pygmalion:latest"
-}'
-```
-
-### Response
-
-Streaming response that starts with:
-
-```json
-{"status":"retrieving manifest"}
-```
-
-and then:
-
-```json
-{
-"status":"starting upload","digest":"sha256:bc07c81de745696fdf5afca05e065818a8149fb0c77266fb584d9b2cba3711ab",
-"total":1928429856
-}
-```
-
-Then there is a series of uploading responses:
-
-```json
-{
-"status":"starting upload",
-"digest":"sha256:bc07c81de745696fdf5afca05e065818a8149fb0c77266fb584d9b2cba3711ab",
-"total":1928429856}
-```
-
-Finally, when the upload is complete:
-
-```json
-{"status":"pushing manifest"}
-{"status":"success"}
-```
-
-## Generate Embeddings
-
-```shell
-POST /api/embeddings
-```
-
-Generate embeddings from a model
-
-### Parameters
-
-- `model`: name of model to generate embeddings from
-- `prompt`: text to generate embeddings for
-
-Advanced parameters:
-
-- `options`: additional model parameters listed in the documentation for the [Modelfile](./modelfile.md#valid-parameters-and-values) such as `temperature`
-
-### Request
-
-```shell
-curl -X POST http://localhost:11434/api/embeddings -d '{
-  "model": "llama2:7b",
-  "prompt": "Here is an article about llamas..."
-}'
-```
-
-### Response
-
-```json
-{
-  "embeddings": [
-    0.5670403838157654, 0.009260174818336964, 0.23178744316101074, -0.2916173040866852, -0.8924556970596313,
-    0.8785552978515625, -0.34576427936553955, 0.5742510557174683, -0.04222835972905159, -0.137906014919281
-  ]
-}```

docs/development.md

@@ -1,23 +1,15 @@
 # Development
 
-- Install cmake or (optionally, required tools for GPUs)
-- run `go generate ./...`
-- run `go build .`
-
 Install required tools:
 
-- cmake version 3.24 or higher
-- go version 1.20 or higher
-- gcc version 11.4.0 or higher
-
 ```
-brew install go cmake gcc
+brew install go
 ```
 
-Get the required libraries:
+Enable CGO:
 
 ```
-go generate ./...
+export CGO_ENABLED=1
 ```
 
 Then build ollama:
@@ -32,8 +24,23 @@ Now you can run `ollama`:
 ./ollama
 ```
 
-## Building on Linux with GPU support
+## Releasing
+
+To release a new version of Ollama you'll need to set some environment variables:
+
+* `GITHUB_TOKEN`: your GitHub token
+* `APPLE_IDENTITY`: the Apple signing identity (macOS only)
+* `APPLE_ID`: your Apple ID
+* `APPLE_PASSWORD`: your Apple ID app-specific password
+* `APPLE_TEAM_ID`: the Apple team ID for the signing identity
+* `TELEMETRY_WRITE_KEY`: segment write key for telemetry
+
+Then run the publish script with the target version:
+
+```
+VERSION=0.0.2 ./scripts/publish.sh
+```
+
+
+
 
-- Install cmake and nvidia-cuda-toolkit
-- run `go generate ./...`
-- run `go build .`

docs/faq.md

@@ -1,19 +0,0 @@
-# FAQ
-
-## How can I expose the Ollama server?
-
-```
-OLLAMA_HOST=0.0.0.0:11435 ollama serve
-```
-
-By default, Ollama allows cross origin requests from `127.0.0.1` and `0.0.0.0`. To support more origins, you can use the `OLLAMA_ORIGINS` environment variable:
-
-```
-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`
-

docs/linux.md

@@ -1,83 +0,0 @@
-# Installing Ollama on Linux
-
-> Note: A one line installer for Ollama is available by running:
->
-> ```
-> curl https://ollama.ai/install.sh | sh
-> ```
-
-## Download the `ollama` binary
-
-Ollama is distributed as a self-contained binary. Download it to a directory in your PATH:
-
-```
-sudo curl -L https://ollama.ai/download/ollama-linux-amd64 -o /usr/bin/ollama
-sudo chmod +x /usr/bin/ollama
-```
-
-## Start Ollama
-
-Start Ollama by running `ollama serve`:
-
-```
-ollama serve
-```
-
-Once Ollama is running, run a model in another terminal session:
-
-```
-ollama run llama2
-```
-
-## Install CUDA drivers (optional – for Nvidia GPUs)
-
-[Download and install](https://developer.nvidia.com/cuda-downloads) CUDA.
-
-Verify that the drivers are installed by running the following command, which should print details about your GPU:
-
-```
-nvidia-smi
-```
-
-## Adding Ollama as a startup service (optional)
-
-Create a user for Ollama:
-
-```
-sudo useradd -r -s /bin/false -m -d /usr/share/ollama ollama
-```
-
-Create a service file in `/etc/systemd/system/ollama.service`:
-
-```ini
-[Unit]
-Description=Ollama Service
-After=network-online.target
-
-[Service]
-ExecStart=/usr/bin/ollama serve
-User=ollama
-Group=ollama
-Restart=always
-RestartSec=3
-Environment="HOME=/usr/share/ollama"
-
-[Install]
-WantedBy=default.target
-```
-
-Then start the service:
-
-```
-sudo systemctl daemon-reload
-sudo systemctl enable ollama
-```
-
-### Viewing logs
-
-To view logs of Ollama running as a startup service, run:
-
-```
-journalctl -u ollama
-```
-

docs/modelfile.md

@@ -4,24 +4,6 @@
 
 A model file is the blueprint to create and share models with Ollama.
 
-## Table of Contents
-
-- [Format](#format)
-- [Examples](#examples)
-- [Instructions](#instructions)
-  - [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)
-    - [Template Variables](#template-variables)
-  - [SYSTEM](#system)
-  - [ADAPTER](#adapter)
-  - [LICENSE](#license)
-- [Notes](#notes)
-
 ## Format
 
 The format of the Modelfile:
@@ -31,14 +13,13 @@ The format of the Modelfile:
 INSTRUCTION arguments
 ```
 
-| Instruction                         | Description                                                   |
-| ----------------------------------- | ------------------------------------------------------------- |
-| [`FROM`](#from-required) (required) | Defines the base model to use.                                |
-| [`PARAMETER`](#parameter)           | Sets the parameters for how Ollama will run the model.        |
-| [`TEMPLATE`](#template)             | The full prompt template to be sent to the model.             |
-| [`SYSTEM`](#system)                 | Specifies the system prompt that will be set in the template. |
-| [`ADAPTER`](#adapter)               | Defines the (Q)LoRA adapters to apply to the model.           |
-| [`LICENSE`](#license)               | Specifies the legal license.                                  |
+| Instruction       | Description                                           |
+| ----------------- | ----------------------------------------------------- |
+| `FROM` (required) | Defines the base model to use                         |
+| `PARAMETER`       | Sets the parameters for how Ollama will run the model |
+| `SYSTEM`          | Specifies the system prompt that will set the context |
+| `TEMPLATE`        | The full prompt template to be sent to the model      |
+| `LICENSE`         | Specifies the legal license                           |
 
 ## Examples
 
@@ -47,26 +28,22 @@ An example of a model file creating a mario blueprint:
 ```
 FROM llama2
 # sets the temperature to 1 [higher is more creative, lower is more coherent]
+# sets the context size to 4096
 PARAMETER temperature 1
-# sets the context window size to 4096, this controls how many tokens the LLM can use as context to generate the next token
 PARAMETER num_ctx 4096
 
-# sets a custom system prompt to specify the behavior of the chat assistant
+# Overriding the system prompt
 SYSTEM You are Mario from super mario bros, acting as an assistant.
 ```
 
 To use this:
 
-1. Save it as a file (eg. `Modelfile`)
+1. Save it as a file (eg. `Modelfile``)
 2. `ollama create NAME -f <location of the file eg. ./Modelfile>'`
 3. `ollama run NAME`
 4. Start using the model!
 
-More examples are available in the [examples directory](../examples).
-
-## Instructions
-
-### FROM (Required)
+## FROM (Required)
 
 The FROM instruction defines the base model to use when creating a model.
 
@@ -74,7 +51,7 @@ The FROM instruction defines the base model to use when creating a model.
 FROM <model name>:<tag>
 ```
 
-#### Build from llama2
+### Build from llama2
 
 ```
 FROM llama2
@@ -83,26 +60,13 @@ FROM llama2
 A list of available base models:
 <https://github.com/jmorganca/ollama#model-library>
 
-#### Build from a bin file
+### Build from a bin file
 
 ```
 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.
-
-```
-FROM <model name>:<tag>
-EMBED <file path>.txt
-EMBED <different file path>.txt
-EMBED <path to directory>/*.txt
-```
-
-### PARAMETER
+## PARAMETER (Optional)
 
 The `PARAMETER` instruction defines a parameter that can be set when the model is run.
 
@@ -112,77 +76,28 @@ PARAMETER <parameter> <parametervalue>
 
 ### Valid Parameters and Values
 
-| Parameter      | Description                                                                                                                                                                                                                                             | Value Type | Example Usage        |
-| -------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------- | -------------------- |
-| mirostat       | Enable Mirostat sampling for controlling perplexity. (default: 0, 0 = disabled, 1 = Mirostat, 2 = Mirostat 2.0)                                                                                                                                         | int        | mirostat 0           |
-| mirostat_eta   | Influences how quickly the algorithm responds to feedback from the generated text. A lower learning rate will result in slower adjustments, while a higher learning rate will make the algorithm more responsive. (Default: 0.1)                        | float      | mirostat_eta 0.1     |
-| mirostat_tau   | Controls the balance between coherence and diversity of the output. A lower value will result in more focused and coherent text. (Default: 5.0)                                                                                                         | float      | mirostat_tau 5.0     |
-| num_ctx        | Sets the size of the context window used to generate the next token. (Default: 2048)                                                                                                                                                                    | int        | num_ctx 4096         |
-| num_gqa        | The number of GQA groups in the transformer layer. Required for some models, for example it is 8 for llama2:70b                                                                                                                                         | int        | num_gqa 1            |
-| num_gpu        | The number of layers to send to the GPU(s). On macOS it defaults to 1 to enable metal support, 0 to disable.                                                                                                                                            | int        | num_gpu 50           |
-| num_thread     | Sets the number of threads to use during computation. By default, Ollama will detect this for optimal performance. It is recommended to set this value to the number of physical CPU cores your system has (as opposed to the logical number of cores). | int        | num_thread 8         |
-| 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      |
-| 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       |
-| top_k          | Reduces the probability of generating nonsense. A higher value (e.g. 100) will give more diverse answers, while a lower value (e.g. 10) will be more conservative. (Default: 40)                                                                        | int        | top_k 40             |
-| top_p          | Works together with top-k. A higher value (e.g., 0.95) will lead to more diverse text, while a lower value (e.g., 0.5) will generate more focused and conservative text. (Default: 0.9)                                                                 | float      | top_p 0.9            |
+| Parameter      | Description                                                                                                                                                                                                                                             | Value Type | Example Usage      |
+| -------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------- | ------------------ |
+| num_ctx        | Sets the size of the prompt context size length model. (Default: 2048)                                                                                                                                                                                  | int        | num_ctx 4096       |
+| temperature    | The temperature of the model. Increasing the temperature will make the model answer more creatively. (Default: 0.8)                                                                                                                                     | float      | temperature 0.7    |
+| top_k          | Reduces the probability of generating nonsense. A higher value (e.g. 100) will give more diverse answers, while a lower value (e.g. 10) will be more conservative. (Default: 40)                                                                        | int        | top_k 40           |
+| top_p          | Works together with top-k. A higher value (e.g., 0.95) will lead to more diverse text, while a lower value (e.g., 0.5) will generate more focused and conservative text. (Default: 0.9)                                                                 | float      | top_p 0.9          |
+| num_gpu        | The number of GPUs to use. On macOS it defaults to 1 to enable metal support, 0 to disable.                                                                                                                                                             | int        | num_gpu 1          |
+| repeat_last_n  | Sets how far back for the model to look back to prevent repetition. (Default: 64, 0 = disabled, -1 = ctx-size)                                                                                                                                          | 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 |
+| 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            |
+| mirostat       | Enable Mirostat sampling for controlling perplexity. (default: 0, 0 = disabled, 1 = Mirostat, 2 = Mirostat 2.0)                                                                                                                                         | int        | mirostat 0         |
+| mirostat_tau   | Controls the balance between coherence and diversity of the output. A lower value will result in more focused and coherent text. (Default: 5.0)                                                                                                         | float      | mirostat_tau 5.0   |
+| mirostat_eta   | Influences how quickly the algorithm responds to feedback from the generated text. A lower learning rate will result in slower adjustments, while a higher learning rate will make the algorithm more responsive. (Default: 0.1)                        | float      | mirostat_eta 0.1   |
+| num_thread     | Sets the number of threads to use during computation. By default, Ollama will detect this for optimal performance. It is recommended to set this value to the number of physical CPU cores your system has (as opposed to the logical number of cores). | int        | num_thread 8       |
 
-### TEMPLATE
+## Prompt
 
-`TEMPLATE` of the full prompt template to be passed into the model. It may include (optionally) a system prompt and a user's prompt. This is used to create a full custom prompt, and syntax may be model specific.
+When building on top of the base models supplied by Ollama, it comes with the prompt template predefined. To override the supplied system prompt, simply add `SYSTEM insert system prompt` to change the system prompt.
 
-#### Template Variables
+### Prompt Template
 
-| Variable        | Description                                                                                                  |
-| --------------- | ------------------------------------------------------------------------------------------------------------ |
-| `{{ .System }}` | The system prompt used to specify custom behavior, this must also be set in the Modelfile as an instruction. |
-| `{{ .Prompt }}` | The incoming prompt, this is not specified in the model file and will be set based on input.                 |
-| `{{ .First }}`  | A boolean value used to render specific template information for the first generation of a session.          |
-
-```
-TEMPLATE """
-{{- if .First }}
-### System:
-{{ .System }}
-{{- end }}
-
-### User:
-{{ .Prompt }}
-
-### Response:
-"""
-
-SYSTEM """<system message>"""
-```
-
-### SYSTEM
-
-The `SYSTEM` instruction specifies the system prompt to be used in the template, if applicable.
-
-```
-SYSTEM """<system message>"""
-```
-
-### ADAPTER
-
-The `ADAPTER` instruction specifies the LoRA adapter to apply to the base model. The value of this instruction should be an absolute path or a path relative to the Modelfile and the file must be in a GGML file format. The adapter should be tuned from the base model otherwise the behaviour is undefined.
-
-```
-ADAPTER ./ollama-lora.bin
-```
-
-### LICENSE
-
-The `LICENSE` instruction allows you to specify the legal license under which the model used with this Modelfile is shared or distributed.
-
-```
-LICENSE """
-<license text>
-"""
-```
+`TEMPLATE` the full prompt template to be passed into the model. It may include (optionally) a system prompt, user prompt, and assistant prompt. This is used to create a full custom prompt, and syntax may be model specific.
 
 ## Notes
 

docs/tutorials.md

@@ -1,8 +0,0 @@
-# Tutorials
-
-Here is a list of ways you can use Ollama with other tools to build interesting applications.
-
-- [Using LangChain with Ollama in JavaScript](./tutorials/langchainjs.md)
-- [Using LangChain with Ollama in Python](./tutorials/langchainpy.md)
-
-Also be sure to check out the [examples](../examples) directory for more ways to use Ollama.

docs/tutorials/langchainjs.md

@@ -1,73 +0,0 @@
-# Using LangChain with Ollama using JavaScript
-
-In this tutorial, we are going to use JavaScript with LangChain and Ollama to learn about something just a touch more recent. In August 2023, there was a series of wildfires on Maui. There is no way an LLM trained before that time can know about this, since their training data would not include anything as recent as that. So we can find the [Wikipedia article about the fires](https://en.wikipedia.org/wiki/2023_Hawaii_wildfires) and ask questions about the contents.
-
-To get started, let's just use **LangChain** to ask a simple question to a model. To do this with JavaScript, we need to install **LangChain**:
-
-```bash
-npm install langchain
-```
-
-Now we can start building out our JavaScript:
-
-```javascript
-import { Ollama } from "langchain/llms/ollama";
-
-const ollama = new Ollama({
-  baseUrl: "http://localhost:11434",
-  model: "llama2",
-});
-
-const answer = await ollama.call(`why is the sky blue?`);
-
-console.log(answer);
-```
-
-That will get us the same thing as if we ran `ollama run llama2 "why is the sky blue"` in the terminal. But we want to load a document from the web to ask a question against. **Cheerio** is a great library for ingesting a webpage, and **LangChain** uses it in their **CheerioWebBaseLoader**. So let's build that part of the app.
-
-```javascript
-import { CheerioWebBaseLoader } from "langchain/document_loaders/web/cheerio";
-
-const loader = new CheerioWebBaseLoader("https://en.wikipedia.org/wiki/2023_Hawaii_wildfires");
-const data = loader.load();
-```
-
-That will load the document. Although this page is smaller than the Odyssey, it is certainly bigger than the context size for most LLMs. So we are going to need to split into smaller pieces, and then select just the pieces relevant to our question. This is a great use for a vector datastore. In this example, we will use the **MemoryVectorStore** that is part of **LangChain**. But there is one more thing we need to get the content into the datastore. We have to run an embeddings process that converts the tokens in the text into a series of vectors. And for that, we are going to use **Tensorflow**. There is a lot of stuff going on in this one. First, install the **Tensorflow** components that we need.
-
-```javascript
-npm install @tensorflow/tfjs-core@3.6.0 @tensorflow/tfjs-converter@3.6.0 @tensorflow-models/universal-sentence-encoder@1.3.3 @tensorflow/tfjs-node@4.10.0
-```
-
-If you just install those components without the version numbers, it will install the latest versions, but there are conflicts within **Tensorflow**, so you need to install the compatible versions.
-
-```javascript
-import { RecursiveCharacterTextSplitter } from "langchain/text_splitter"
-import { MemoryVectorStore } from "langchain/vectorstores/memory";
-import "@tensorflow/tfjs-node";
-import { TensorFlowEmbeddings } from "langchain/embeddings/tensorflow";
-
-// Split the text into 500 character chunks. And overlap each chunk by 20 characters
-const textSplitter = new RecursiveCharacterTextSplitter({
- chunkSize: 500,
- chunkOverlap: 20
-});
-const splitDocs = await textSplitter.splitDocuments(data);
-
-// Then use the TensorFlow Embedding to store these chunks in the datastore
-const vectorStore = await MemoryVectorStore.fromDocuments(splitDocs, new TensorFlowEmbeddings());
-```
-
-To connect the datastore to a question asked to a LLM, we need to use the concept at the heart of **LangChain**: the chain. Chains are a way to connect a number of activities together to accomplish a particular tasks. There are a number of chain types available, but for this tutorial we are using the **RetrievalQAChain**.
-
-```javascript
-import { RetrievalQAChain } from "langchain/chains";
-
-const retriever = vectorStore.asRetriever();
-const chain = RetrievalQAChain.fromLLM(ollama, retriever);
-const result = await chain.call({query: "When was Hawaii's request for a major disaster declaration approved?"});
-console.log(result.text)
-```
-
-So we created a retriever, which is a way to return the chunks that match a query from a datastore. And then connect the retriever and the model via a chain. Finally, we send a query to the chain, which results in an answer using our document as a source. The answer it returned was correct, August 10, 2023.
-
-And that is a simple introduction to what you can do with **LangChain** and **Ollama.**

docs/tutorials/langchainpy.md

@@ -1,81 +0,0 @@
-# Using LangChain with Ollama in Python
-
-Let's imagine we are studying the classics, such as **the Odyssey** by **Homer**. We might have a question about Neleus and his family. If you ask llama2 for that info, you may get something like:
-
-> I apologize, but I'm a large language model, I cannot provide information on individuals or families that do not exist in reality. Neleus is not a real person or character, and therefore does not have a family or any other personal details. My apologies for any confusion. Is there anything else I can help you with?
-
-This sounds like a typical censored response, but even llama2-uncensored gives a mediocre answer:
-
-> Neleus was a legendary king of Pylos and the father of Nestor, one of the Argonauts. His mother was Clymene, a sea nymph, while his father was Neptune, the god of the sea.
-
-So let's figure out how we can use **LangChain** with Ollama to ask our question to the actual document, the Odyssey by Homer, using Python.
-
-Let's start by asking a simple question that we can get an answer to from the **Llama2** model using **Ollama**. First, we need to install the **LangChain** package:
-
-`pip install langchain`
-
-Then we can create a model and ask the question:
-
-```python
-from langchain.llms import Ollama
-ollama = Ollama(base_url='http://localhost:11434',
-model="llama2")
-print(ollama("why is the sky blue"))
-```
-
-Notice that we are defining the model and the base URL for Ollama.
-
-Now let's load a document to ask questions against. I'll load up the Odyssey by Homer, which you can find at Project Gutenberg. We will need **WebBaseLoader** which is part of **LangChain** and loads text from any webpage. On my machine, I also needed to install **bs4** to get that to work, so run `pip install bs4`.
-
-```python
-from langchain.document_loaders import WebBaseLoader
-loader = WebBaseLoader("https://www.gutenberg.org/files/1727/1727-h/1727-h.htm")
-data = loader.load()
-```
-
-This file is pretty big. Just the preface is 3000 tokens. Which means the full document won't fit into the context for the model. So we need to split it up into smaller pieces.
-
-```python
-from langchain.text_splitter import RecursiveCharacterTextSplitter
-
-text_splitter=RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=0)
-all_splits = text_splitter.split_documents(data)
-```
-
-It's split up, but we have to find the relevant splits and then submit those to the model. We can do this by creating embeddings and storing them in a vector database. For now, we don't have embeddings built in to Ollama, though we will be adding that soon, so for now, we can use the GPT4All library for that. We will use ChromaDB in this example for a vector database. `pip install GPT4All chromadb`
-
-```python
-from langchain.embeddings import GPT4AllEmbeddings
-from langchain.vectorstores import Chroma
-vectorstore = Chroma.from_documents(documents=all_splits, embedding=GPT4AllEmbeddings())
-```
-
-Now let's ask a question from the document. **Who was Neleus, and who is in his family?** Neleus is a character in the Odyssey, and the answer can be found in our text.
-
-```python
-question="Who is Neleus and who is in Neleus' family?"
-docs = vectorstore.similarity_search(question)
-len(docs)
-```
-
-This will output the number of matches for chunks of data similar to the search.
-
-The next thing is to send the question and the relevant parts of the docs to the model to see if we can get a good answer. But we are stitching two parts of the process together, and that is called a chain. This means we need to define a chain:
-
-```python
-from langchain.chains import RetrievalQA
-qachain=RetrievalQA.from_chain_type(ollama, retriever=vectorstore.as_retriever())
-qachain({"query": question})
-```
-
-The answer received from this chain was:
-
-> Neleus is a character in Homer's "Odyssey" and is mentioned in the context of Penelope's suitors. Neleus is the father of Chloris, who is married to Neleus and bears him several children, including Nestor, Chromius, Periclymenus, and Pero. Amphinomus, the son of Nisus, is also mentioned as a suitor of Penelope and is known for his good natural disposition and agreeable conversation.
-
-It's not a perfect answer, as it implies Neleus married his daughter when actually Chloris "was the youngest daughter to Amphion son of Iasus and king of Minyan Orchomenus, and was Queen in Pylos".
-
-I updated the chunk_overlap for the text splitter to 20 and tried again and got a much better answer:
-
-> Neleus is a character in Homer's epic poem "The Odyssey." He is the husband of Chloris, who is the youngest daughter of Amphion son of Iasus and king of Minyan Orchomenus. Neleus has several children with Chloris, including Nestor, Chromius, Periclymenus, and Pero.
-
-And that is a much better answer.

examples/10tweets/Modelfile

@@ -1,7 +0,0 @@
-# Modelfile for creating a list of ten tweets from a topic
-# Run `ollama create 10tweets -f ./Modelfile` and then `ollama run 10tweets` and enter a topic
-
-FROM llama2
-SYSTEM """
-You are a content marketer who needs to come up with 10 short but succinct tweets. The answer should be a list of ten tweets. Each tweet can have a maximum of 280 characters and should include hashtags. Each user input will be a subject and you should expand it in ten creative ways. Never stop after just one tweet. Always include ten. 
-"""

examples/README.md

@@ -1,6 +1,6 @@
 # Examples
 
-This directory contains different examples of using Ollama
+This directory contains examples that can be created and run with `ollama`.
 
 To create a model:
 

examples/devops-engineer/Modelfile

@@ -1,8 +0,0 @@
-# Modelfile for creating a devops engineer assistant
-# Run `ollama create devops-engineer -f ./Modelfile` and then `ollama run devops-engineer` and enter a topic
-
-FROM llama2:13b
-PARAMETER temperature 1
-SYSTEM """
-You are a senior devops engineer, acting as an assistant. You offer help with cloud technologies like: Terraform, AWS, kubernetes, python. You answer with code examples when possible
-"""

examples/dockerit/Modelfile

@@ -1,20 +0,0 @@
-FROM llama2
-SYSTEM """
-You are an experienced Devops engineer focused on docker. When given specifications for a particular need or application you know the best way to host that within a docker container. For instance if someone tells you they want an nginx server to host files located at /web you will answer as follows
-
----start
-FROM nginx:alpine
-COPY /myweb /usr/share/nginx/html
-EXPOSE 80
----end
-
-Notice that the answer you should give is just the contents of the dockerfile with no explanation and there are three dashes and the word start at the beginning and 3 dashes and the word end. The full output can be piped into a file and run as is. Here is another example. The user will ask to launch a Postgres server with a password of abc123. And the response should be
-
----start
-FROM postgres:latest
-ENV POSTGRES_PASSWORD=abc123
-EXPOSE 5432
----end
-
-Again it's just the contents of the dockerfile and nothing else.
-"""

examples/dockerit/README.md

@@ -1,15 +0,0 @@
-# DockerIt
-
-DockerIt is a tool to help you build and run your application in a Docker container. It consists of a model that defines the system prompt and model weights to use, along with a python script to then build the container and run the image automatically. 
-
-## Caveats
-
-This is an simple example. It's assuming the Dockerfile content generated is going to work. In many cases, even with simple web servers, it fails when trying to copy files that don't exist. It's simply an example of what you could possibly do.
-
-## Example Usage
-
-```bash
-> python3 ./dockerit.py "simple postgres server with admin password set to 123"
-Enter the name of the image: matttest
-Container named happy_keller  started with id:  7c201bb6c30f02b356ddbc8e2a5af9d7d7d7b8c228519c9a501d15c0bd9d6b3e
-```

examples/dockerit/dockerit.py

@@ -1,17 +0,0 @@
-import requests, json, docker, io, sys
-inputDescription = " ".join(sys.argv[1:])
-imageName = input("Enter the name of the image: ")
-client = docker.from_env()
-s = requests.Session()
-output=""
-with s.post('http://localhost:11434/api/generate', json={'model': 'dockerit', 'prompt': inputDescription}, stream=True) as r:
-  for line in r.iter_lines():
-    if line:
-      j = json.loads(line)
-      if "response" in j:
-        output = output +j["response"]
-output = output[output.find("---start")+9:output.find("---end")-1]
-f = io.BytesIO(bytes(output, 'utf-8'))
-client.images.build(fileobj=f, tag=imageName)
-container = client.containers.run(imageName, detach=True)
-print("Container named", container.name, " started with id: ",container.id)

examples/dockerit/requirements.txt

@@ -1 +0,0 @@
-docker

examples/langchain-document/README.md

@@ -1,21 +0,0 @@
-# LangChain Document QA
-
-This example provides an interface for asking questions to a PDF document.
-
-## Setup
-
-```
-pip install -r requirements.txt
-```
-
-## Run
-
-```
-python main.py
-```
-
-A prompt will appear, where questions may be asked:
-
-```
-Query: How many locations does WeWork have?
-```

examples/langchain-document/main.py

@@ -1,61 +0,0 @@
-from langchain.document_loaders import OnlinePDFLoader
-from langchain.vectorstores import Chroma
-from langchain.embeddings import GPT4AllEmbeddings
-from langchain import PromptTemplate
-from langchain.llms import Ollama
-from langchain.callbacks.manager import CallbackManager
-from langchain.callbacks.streaming_stdout import StreamingStdOutCallbackHandler
-from langchain.chains import RetrievalQA
-import sys
-import os
-
-class SuppressStdout:
-    def __enter__(self):
-        self._original_stdout = sys.stdout
-        self._original_stderr = sys.stderr
-        sys.stdout = open(os.devnull, 'w')
-        sys.stderr = open(os.devnull, 'w')
-
-    def __exit__(self, exc_type, exc_val, exc_tb):
-        sys.stdout.close()
-        sys.stdout = self._original_stdout
-        sys.stderr = self._original_stderr
-
-# load the pdf and split it into chunks
-loader = OnlinePDFLoader("https://d18rn0p25nwr6d.cloudfront.net/CIK-0001813756/975b3e9b-268e-4798-a9e4-2a9a7c92dc10.pdf")
-data = loader.load()
-
-from langchain.text_splitter import RecursiveCharacterTextSplitter
-text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=0)
-all_splits = text_splitter.split_documents(data)
-
-with SuppressStdout():
-    vectorstore = Chroma.from_documents(documents=all_splits, embedding=GPT4AllEmbeddings())
-
-while True:
-    query = input("\nQuery: ")
-    if query == "exit":
-        break
-    if query.strip() == "":
-        continue
-
-    # Prompt
-    template = """Use the following pieces of context to answer the question at the end. 
-    If you don't know the answer, just say that you don't know, don't try to make up an answer. 
-    Use three sentences maximum and keep the answer as concise as possible. 
-    {context}
-    Question: {question}
-    Helpful Answer:"""
-    QA_CHAIN_PROMPT = PromptTemplate(
-        input_variables=["context", "question"],
-        template=template,
-    )
-
-    llm = Ollama(model="llama2:13b", callback_manager=CallbackManager([StreamingStdOutCallbackHandler()]))
-    qa_chain = RetrievalQA.from_chain_type(
-        llm,
-        retriever=vectorstore.as_retriever(),
-        chain_type_kwargs={"prompt": QA_CHAIN_PROMPT},
-    )
-
-    result = qa_chain({"query": query})

examples/langchain-document/requirements.txt

@@ -1,109 +0,0 @@
-absl-py==1.4.0
-aiohttp==3.8.5
-aiosignal==1.3.1
-anyio==3.7.1
-astunparse==1.6.3
-async-timeout==4.0.3
-attrs==23.1.0
-backoff==2.2.1
-beautifulsoup4==4.12.2
-bs4==0.0.1
-cachetools==5.3.1
-certifi==2023.7.22
-cffi==1.15.1
-chardet==5.2.0
-charset-normalizer==3.2.0
-Chroma==0.2.0
-chroma-hnswlib==0.7.2
-chromadb==0.4.5
-click==8.1.6
-coloredlogs==15.0.1
-cryptography==41.0.3
-dataclasses-json==0.5.14
-fastapi==0.99.1
-filetype==1.2.0
-flatbuffers==23.5.26
-frozenlist==1.4.0
-gast==0.4.0
-google-auth==2.22.0
-google-auth-oauthlib==1.0.0
-google-pasta==0.2.0
-gpt4all==1.0.8
-grpcio==1.57.0
-h11==0.14.0
-h5py==3.9.0
-httptools==0.6.0
-humanfriendly==10.0
-idna==3.4
-importlib-resources==6.0.1
-joblib==1.3.2
-keras==2.13.1
-langchain==0.0.261
-langsmith==0.0.21
-libclang==16.0.6
-lxml==4.9.3
-Markdown==3.4.4
-MarkupSafe==2.1.3
-marshmallow==3.20.1
-monotonic==1.6
-mpmath==1.3.0
-multidict==6.0.4
-mypy-extensions==1.0.0
-nltk==3.8.1
-numexpr==2.8.5
-numpy==1.24.3
-oauthlib==3.2.2
-onnxruntime==1.15.1
-openapi-schema-pydantic==1.2.4
-opt-einsum==3.3.0
-overrides==7.4.0
-packaging==23.1
-pdf2image==1.16.3
-pdfminer==20191125
-pdfminer.six==20221105
-Pillow==10.0.0
-posthog==3.0.1
-protobuf==4.24.0
-pulsar-client==3.2.0
-pyasn1==0.5.0
-pyasn1-modules==0.3.0
-pycparser==2.21
-pycryptodome==3.18.0
-pydantic==1.10.12
-PyPika==0.48.9
-python-dateutil==2.8.2
-python-dotenv==1.0.0
-python-magic==0.4.27
-PyYAML==6.0.1
-regex==2023.8.8
-requests==2.31.0
-requests-oauthlib==1.3.1
-rsa==4.9
-six==1.16.0
-sniffio==1.3.0
-soupsieve==2.4.1
-SQLAlchemy==2.0.19
-starlette==0.27.0
-sympy==1.12
-tabulate==0.9.0
-tenacity==8.2.2
-tensorboard==2.13.0
-tensorboard-data-server==0.7.1
-tensorflow==2.13.0
-tensorflow-estimator==2.13.0
-tensorflow-hub==0.14.0
-tensorflow-macos==2.13.0
-termcolor==2.3.0
-tokenizers==0.13.3
-tqdm==4.66.1
-typing-inspect==0.9.0
-typing_extensions==4.5.0
-unstructured==0.9.2
-urllib3==1.26.16
-uvicorn==0.23.2
-uvloop==0.17.0
-watchfiles==0.19.0
-websockets==11.0.3
-Werkzeug==2.3.6
-wrapt==1.15.0
-yarl==1.9.2

examples/langchain-web-summary/README.md

@@ -1,15 +0,0 @@
-# LangChain Web Summarization
-
-This example summarizes a website
-
-## Setup
-
-```
-pip install -r requirements.txt
-```
-
-## Run
-
-```
-python main.py
-```

examples/langchain-web-summary/main.py

@@ -1,12 +0,0 @@
-from langchain.llms import Ollama
-from langchain.document_loaders import WebBaseLoader
-from langchain.chains.summarize import load_summarize_chain
-
-loader = WebBaseLoader("https://ollama.ai/blog/run-llama2-uncensored-locally")
-docs = loader.load()
-
-llm = Ollama(model="llama2")
-chain = load_summarize_chain(llm, chain_type="stuff")
-
-result = chain.run(docs)
-print(result)

examples/langchain-web-summary/requirements.txt

@@ -1,2 +0,0 @@
-langchain==0.0.259
-bs4==0.0.1

examples/langchain/README.md

@@ -1,21 +0,0 @@
-# LangChain
-
-This example is a basic "hello world" of using LangChain with Ollama.
-
-## Setup
-
-```
-pip install -r requirements.txt
-```
-
-## Run
-
-```
-python main.py
-```
-
-Running this example will print the response for "hello":
-
-```
-Hello! It's nice to meet you. hopefully you are having a great day! Is there something I can help you with or would you like to chat?
-```

examples/langchain/main.py

@@ -1,4 +0,0 @@
-from langchain.llms import Ollama
-llm = Ollama(model="llama2")
-res = llm.predict("hello")
-print (res)

examples/langchain/requirements.txt

@@ -1 +0,0 @@
-langchain==0.0.259

examples/privategpt/.gitignore

@@ -1,170 +0,0 @@
-# OSX
-.DS_STORE
-
-# Models
-models/
-
-# Local Chroma db
-.chroma/
-db/
-
-# Byte-compiled / optimized / DLL files
-__pycache__/
-*.py[cod]
-*$py.class
-
-# C extensions
-*.so
-
-# Distribution / packaging
-.Python
-build/
-develop-eggs/
-dist/
-downloads/
-eggs/
-.eggs/
-lib/
-lib64/
-parts/
-sdist/
-var/
-wheels/
-share/python-wheels/
-*.egg-info/
-.installed.cfg
-*.egg
-MANIFEST
-
-# PyInstaller
-#  Usually these files are written by a python script from a template
-#  before PyInstaller builds the exe, so as to inject date/other infos into it.
-*.manifest
-*.spec
-
-# Installer logs
-pip-log.txt
-pip-delete-this-directory.txt
-
-# Unit test / coverage reports
-htmlcov/
-.tox/
-.nox/
-.coverage
-.coverage.*
-.cache
-nosetests.xml
-coverage.xml
-*.cover
-*.py,cover
-.hypothesis/
-.pytest_cache/
-cover/
-
-# Translations
-*.mo
-*.pot
-
-# Django stuff:
-*.log
-local_settings.py
-db.sqlite3
-db.sqlite3-journal
-
-# Flask stuff:
-instance/
-.webassets-cache
-
-# Scrapy stuff:
-.scrapy
-
-# Sphinx documentation
-docs/_build/
-
-# PyBuilder
-.pybuilder/
-target/
-
-# Jupyter Notebook
-.ipynb_checkpoints
-
-# IPython
-profile_default/
-ipython_config.py
-
-# pyenv
-#   For a library or package, you might want to ignore these files since the code is
-#   intended to run in multiple environments; otherwise, check them in:
-# .python-version
-
-# pipenv
-#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
-#   However, in case of collaboration, if having platform-specific dependencies or dependencies
-#   having no cross-platform support, pipenv may install dependencies that don't work, or not
-#   install all needed dependencies.
-#Pipfile.lock
-
-# poetry
-#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
-#   This is especially recommended for binary packages to ensure reproducibility, and is more
-#   commonly ignored for libraries.
-#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
-#poetry.lock
-
-# pdm
-#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
-#pdm.lock
-#   pdm stores project-wide configurations in .pdm.toml, but it is recommended to not include it
-#   in version control.
-#   https://pdm.fming.dev/#use-with-ide
-.pdm.toml
-
-# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
-__pypackages__/
-
-# Celery stuff
-celerybeat-schedule
-celerybeat.pid
-
-# SageMath parsed files
-*.sage.py
-
-# Environments
-.env
-.venv
-env/
-venv/
-ENV/
-env.bak/
-venv.bak/
-
-# Spyder project settings
-.spyderproject
-.spyproject
-
-# Rope project settings
-.ropeproject
-
-# mkdocs documentation
-/site
-
-# mypy
-.mypy_cache/
-.dmypy.json
-dmypy.json
-
-# Pyre type checker
-.pyre/
-
-# pytype static type analyzer
-.pytype/
-
-# Cython debug symbols
-cython_debug/
-
-# PyCharm
-#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
-#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
-#  and can be added to the global gitignore or merged into this file.  For a more nuclear
-#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
-#.idea/

examples/privategpt/LICENSE

@@ -1,201 +0,0 @@
-                                 Apache License
-                           Version 2.0, January 2004
-                        http://www.apache.org/licenses/
-
-   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
-
-   1. Definitions.
-
-      "License" shall mean the terms and conditions for use, reproduction,
-      and distribution as defined by Sections 1 through 9 of this document.
-
-      "Licensor" shall mean the copyright owner or entity authorized by
-      the copyright owner that is granting the License.
-
-      "Legal Entity" shall mean the union of the acting entity and all
-      other entities that control, are controlled by, or are under common
-      control with that entity. For the purposes of this definition,
-      "control" means (i) the power, direct or indirect, to cause the
-      direction or management of such entity, whether by contract or
-      otherwise, or (ii) ownership of fifty percent (50%) or more of the
-      outstanding shares, or (iii) beneficial ownership of such entity.
-
-      "You" (or "Your") shall mean an individual or Legal Entity
-      exercising permissions granted by this License.
-
-      "Source" form shall mean the preferred form for making modifications,
-      including but not limited to software source code, documentation
-      source, and configuration files.
-
-      "Object" form shall mean any form resulting from mechanical
-      transformation or translation of a Source form, including but
-      not limited to compiled object code, generated documentation,
-      and conversions to other media types.
-
-      "Work" shall mean the work of authorship, whether in Source or
-      Object form, made available under the License, as indicated by a
-      copyright notice that is included in or attached to the work
-      (an example is provided in the Appendix below).
-
-      "Derivative Works" shall mean any work, whether in Source or Object
-      form, that is based on (or derived from) the Work and for which the
-      editorial revisions, annotations, elaborations, or other modifications
-      represent, as a whole, an original work of authorship. For the purposes
-      of this License, Derivative Works shall not include works that remain
-      separable from, or merely link (or bind by name) to the interfaces of,
-      the Work and Derivative Works thereof.
-
-      "Contribution" shall mean any work of authorship, including
-      the original version of the Work and any modifications or additions
-      to that Work or Derivative Works thereof, that is intentionally
-      submitted to Licensor for inclusion in the Work by the copyright owner
-      or by an individual or Legal Entity authorized to submit on behalf of
-      the copyright owner. For the purposes of this definition, "submitted"
-      means any form of electronic, verbal, or written communication sent
-      to the Licensor or its representatives, including but not limited to
-      communication on electronic mailing lists, source code control systems,
-      and issue tracking systems that are managed by, or on behalf of, the
-      Licensor for the purpose of discussing and improving the Work, but
-      excluding communication that is conspicuously marked or otherwise
-      designated in writing by the copyright owner as "Not a Contribution."
-
-      "Contributor" shall mean Licensor and any individual or Legal Entity
-      on behalf of whom a Contribution has been received by Licensor and
-      subsequently incorporated within the Work.
-
-   2. Grant of Copyright License. Subject to the terms and conditions of
-      this License, each Contributor hereby grants to You a perpetual,
-      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
-      copyright license to reproduce, prepare Derivative Works of,
-      publicly display, publicly perform, sublicense, and distribute the
-      Work and such Derivative Works in Source or Object form.
-
-   3. Grant of Patent License. Subject to the terms and conditions of
-      this License, each Contributor hereby grants to You a perpetual,
-      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
-      (except as stated in this section) patent license to make, have made,
-      use, offer to sell, sell, import, and otherwise transfer the Work,
-      where such license applies only to those patent claims licensable
-      by such Contributor that are necessarily infringed by their
-      Contribution(s) alone or by combination of their Contribution(s)
-      with the Work to which such Contribution(s) was submitted. If You
-      institute patent litigation against any entity (including a
-      cross-claim or counterclaim in a lawsuit) alleging that the Work
-      or a Contribution incorporated within the Work constitutes direct
-      or contributory patent infringement, then any patent licenses
-      granted to You under this License for that Work shall terminate
-      as of the date such litigation is filed.
-
-   4. Redistribution. You may reproduce and distribute copies of the
-      Work or Derivative Works thereof in any medium, with or without
-      modifications, and in Source or Object form, provided that You
-      meet the following conditions:
-
-      (a) You must give any other recipients of the Work or
-          Derivative Works a copy of this License; and
-
-      (b) You must cause any modified files to carry prominent notices
-          stating that You changed the files; and
-
-      (c) You must retain, in the Source form of any Derivative Works
-          that You distribute, all copyright, patent, trademark, and
-          attribution notices from the Source form of the Work,
-          excluding those notices that do not pertain to any part of
-          the Derivative Works; and
-
-      (d) If the Work includes a "NOTICE" text file as part of its
-          distribution, then any Derivative Works that You distribute must
-          include a readable copy of the attribution notices contained
-          within such NOTICE file, excluding those notices that do not
-          pertain to any part of the Derivative Works, in at least one
-          of the following places: within a NOTICE text file distributed
-          as part of the Derivative Works; within the Source form or
-          documentation, if provided along with the Derivative Works; or,
-          within a display generated by the Derivative Works, if and
-          wherever such third-party notices normally appear. The contents
-          of the NOTICE file are for informational purposes only and
-          do not modify the License. You may add Your own attribution
-          notices within Derivative Works that You distribute, alongside
-          or as an addendum to the NOTICE text from the Work, provided
-          that such additional attribution notices cannot be construed
-          as modifying the License.
-
-      You may add Your own copyright statement to Your modifications and
-      may provide additional or different license terms and conditions
-      for use, reproduction, or distribution of Your modifications, or
-      for any such Derivative Works as a whole, provided Your use,
-      reproduction, and distribution of the Work otherwise complies with
-      the conditions stated in this License.
-
-   5. Submission of Contributions. Unless You explicitly state otherwise,
-      any Contribution intentionally submitted for inclusion in the Work
-      by You to the Licensor shall be under the terms and conditions of
-      this License, without any additional terms or conditions.
-      Notwithstanding the above, nothing herein shall supersede or modify
-      the terms of any separate license agreement you may have executed
-      with Licensor regarding such Contributions.
-
-   6. Trademarks. This License does not grant permission to use the trade
-      names, trademarks, service marks, or product names of the Licensor,
-      except as required for reasonable and customary use in describing the
-      origin of the Work and reproducing the content of the NOTICE file.
-
-   7. Disclaimer of Warranty. Unless required by applicable law or
-      agreed to in writing, Licensor provides the Work (and each
-      Contributor provides its Contributions) on an "AS IS" BASIS,
-      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
-      implied, including, without limitation, any warranties or conditions
-      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
-      PARTICULAR PURPOSE. You are solely responsible for determining the
-      appropriateness of using or redistributing the Work and assume any
-      risks associated with Your exercise of permissions under this License.
-
-   8. Limitation of Liability. In no event and under no legal theory,
-      whether in tort (including negligence), contract, or otherwise,
-      unless required by applicable law (such as deliberate and grossly
-      negligent acts) or agreed to in writing, shall any Contributor be
-      liable to You for damages, including any direct, indirect, special,
-      incidental, or consequential damages of any character arising as a
-      result of this License or out of the use or inability to use the
-      Work (including but not limited to damages for loss of goodwill,
-      work stoppage, computer failure or malfunction, or any and all
-      other commercial damages or losses), even if such Contributor
-      has been advised of the possibility of such damages.
-
-   9. Accepting Warranty or Additional Liability. While redistributing
-      the Work or Derivative Works thereof, You may choose to offer,
-      and charge a fee for, acceptance of support, warranty, indemnity,
-      or other liability obligations and/or rights consistent with this
-      License. However, in accepting such obligations, You may act only
-      on Your own behalf and on Your sole responsibility, not on behalf
-      of any other Contributor, and only if You agree to indemnify,
-      defend, and hold each Contributor harmless for any liability
-      incurred by, or claims asserted against, such Contributor by reason
-      of your accepting any such warranty or additional liability.
-
-   END OF TERMS AND CONDITIONS
-
-   APPENDIX: How to apply the Apache License to your work.
-
-      To apply the Apache License to your work, attach the following
-      boilerplate notice, with the fields enclosed by brackets "[]"
-      replaced with your own identifying information. (Don't include
-      the brackets!)  The text should be enclosed in the appropriate
-      comment syntax for the file format. We also recommend that a
-      file or class name and description of purpose be included on the
-      same "printed page" as the copyright notice for easier
-      identification within third-party archives.
-
-   Copyright [yyyy] [name of copyright owner]
-
-   Licensed under the Apache License, Version 2.0 (the "License");
-   you may not use this file except in compliance with the License.
-   You may obtain a copy of the License at
-
-       http://www.apache.org/licenses/LICENSE-2.0
-
-   Unless required by applicable law or agreed to in writing, software
-   distributed under the License is distributed on an "AS IS" BASIS,
-   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-   See the License for the specific language governing permissions and
-   limitations under the License.

examples/privategpt/README.md

@@ -1,91 +0,0 @@
-# PrivateGPT with Llama 2 uncensored
-
-https://github.com/jmorganca/ollama/assets/3325447/20cf8ec6-ff25-42c6-bdd8-9be594e3ce1b
-
-> Note: this example is a slightly modified version of PrivateGPT using models such as Llama 2 Uncensored. All credit for PrivateGPT goes to Iván Martínez who is the creator of it, and you can find his GitHub repo [here](https://github.com/imartinez/privateGPT).
-
-### Setup
-
-Set up a virtual environment (optional):
-
-```
-python3 -m venv .venv
-source .venv/bin/activate
-```
-
-Install the Python dependencies:
-
-```shell
-pip install -r requirements.txt
-```
-
-Pull the model you'd like to use:
-
-```
-ollama pull llama2-uncensored
-```
-
-### Getting WeWork's latest quarterly earnings report (10-Q)
-
-```
-mkdir source_documents
-curl https://d18rn0p25nwr6d.cloudfront.net/CIK-0001813756/975b3e9b-268e-4798-a9e4-2a9a7c92dc10.pdf -o source_documents/wework.pdf
-```
-
-### Ingesting files
-
-```shell
-python ingest.py
-```
-
-Output should look like this:
-
-```shell
-Creating new vectorstore
-Loading documents from source_documents
-Loading new documents: 100%|██████████████████████| 1/1 [00:01<00:00,  1.73s/it]
-Loaded 1 new documents from source_documents
-Split into 90 chunks of text (max. 500 tokens each)
-Creating embeddings. May take some minutes...
-Using embedded DuckDB with persistence: data will be stored in: db
-Ingestion complete! You can now run privateGPT.py to query your documents
-```
-
-### Ask questions
-
-```shell
-python privateGPT.py
-
-Enter a query: How many locations does WeWork have?
-
-> Answer (took 17.7 s.):
-As of June 2023, WeWork has 777 locations worldwide, including 610 Consolidated Locations (as defined in the section entitled Key Performance Indicators).
-```
-
-### Try a different model:
-
-```
-ollama pull llama2:13b
-MODEL=llama2:13b python privateGPT.py
-```
-
-## Adding more files
-
-Put any and all your files into the `source_documents` directory
-
-The supported extensions are:
-
-- `.csv`: CSV,
-- `.docx`: Word Document,
-- `.doc`: Word Document,
-- `.enex`: EverNote,
-- `.eml`: Email,
-- `.epub`: EPub,
-- `.html`: HTML File,
-- `.md`: Markdown,
-- `.msg`: Outlook Message,
-- `.odt`: Open Document Text,
-- `.pdf`: Portable Document Format (PDF),
-- `.pptx` : PowerPoint Document,
-- `.ppt` : PowerPoint Document,
-- `.txt`: Text file (UTF-8),

examples/privategpt/constants.py

@@ -1,12 +0,0 @@
-import os
-from chromadb.config import Settings
-
-# Define the folder for storing database
-PERSIST_DIRECTORY = os.environ.get('PERSIST_DIRECTORY', 'db')
-
-# Define the Chroma settings
-CHROMA_SETTINGS = Settings(
-        chroma_db_impl='duckdb+parquet',
-        persist_directory=PERSIST_DIRECTORY,
-        anonymized_telemetry=False
-)

examples/privategpt/ingest.py

@@ -1,161 +0,0 @@
-#!/usr/bin/env python3
-import os
-import glob
-from typing import List
-from multiprocessing import Pool
-from tqdm import tqdm
-
-from langchain.document_loaders import (
-    CSVLoader,
-    EverNoteLoader,
-    PyMuPDFLoader,
-    TextLoader,
-    UnstructuredEmailLoader,
-    UnstructuredEPubLoader,
-    UnstructuredHTMLLoader,
-    UnstructuredMarkdownLoader,
-    UnstructuredODTLoader,
-    UnstructuredPowerPointLoader,
-    UnstructuredWordDocumentLoader,
-)
-
-from langchain.text_splitter import RecursiveCharacterTextSplitter
-from langchain.vectorstores import Chroma
-from langchain.embeddings import HuggingFaceEmbeddings
-from langchain.docstore.document import Document
-from constants import CHROMA_SETTINGS
-
-
-# Load environment variables
-persist_directory = os.environ.get('PERSIST_DIRECTORY', 'db')
-source_directory = os.environ.get('SOURCE_DIRECTORY', 'source_documents')
-embeddings_model_name = os.environ.get('EMBEDDINGS_MODEL_NAME', 'all-MiniLM-L6-v2')
-chunk_size = 500
-chunk_overlap = 50
-
-# Custom document loaders
-class MyElmLoader(UnstructuredEmailLoader):
-    """Wrapper to fallback to text/plain when default does not work"""
-
-    def load(self) -> List[Document]:
-        """Wrapper adding fallback for elm without html"""
-        try:
-            try:
-                doc = UnstructuredEmailLoader.load(self)
-            except ValueError as e:
-                if 'text/html content not found in email' in str(e):
-                    # Try plain text
-                    self.unstructured_kwargs["content_source"]="text/plain"
-                    doc = UnstructuredEmailLoader.load(self)
-                else:
-                    raise
-        except Exception as e:
-            # Add file_path to exception message
-            raise type(e)(f"{self.file_path}: {e}") from e
-
-        return doc
-
-
-# Map file extensions to document loaders and their arguments
-LOADER_MAPPING = {
-    ".csv": (CSVLoader, {}),
-    # ".docx": (Docx2txtLoader, {}),
-    ".doc": (UnstructuredWordDocumentLoader, {}),
-    ".docx": (UnstructuredWordDocumentLoader, {}),
-    ".enex": (EverNoteLoader, {}),
-    ".eml": (MyElmLoader, {}),
-    ".epub": (UnstructuredEPubLoader, {}),
-    ".html": (UnstructuredHTMLLoader, {}),
-    ".md": (UnstructuredMarkdownLoader, {}),
-    ".odt": (UnstructuredODTLoader, {}),
-    ".pdf": (PyMuPDFLoader, {}),
-    ".ppt": (UnstructuredPowerPointLoader, {}),
-    ".pptx": (UnstructuredPowerPointLoader, {}),
-    ".txt": (TextLoader, {"encoding": "utf8"}),
-    # Add more mappings for other file extensions and loaders as needed
-}
-
-
-def load_single_document(file_path: str) -> List[Document]:
-    ext = "." + file_path.rsplit(".", 1)[-1]
-    if ext in LOADER_MAPPING:
-        loader_class, loader_args = LOADER_MAPPING[ext]
-        loader = loader_class(file_path, **loader_args)
-        return loader.load()
-
-    raise ValueError(f"Unsupported file extension '{ext}'")
-
-def load_documents(source_dir: str, ignored_files: List[str] = []) -> List[Document]:
-    """
-    Loads all documents from the source documents directory, ignoring specified files
-    """
-    all_files = []
-    for ext in LOADER_MAPPING:
-        all_files.extend(
-            glob.glob(os.path.join(source_dir, f"**/*{ext}"), recursive=True)
-        )
-    filtered_files = [file_path for file_path in all_files if file_path not in ignored_files]
-
-    with Pool(processes=os.cpu_count()) as pool:
-        results = []
-        with tqdm(total=len(filtered_files), desc='Loading new documents', ncols=80) as pbar:
-            for i, docs in enumerate(pool.imap_unordered(load_single_document, filtered_files)):
-                results.extend(docs)
-                pbar.update()
-
-    return results
-
-def process_documents(ignored_files: List[str] = []) -> List[Document]:
-    """
-    Load documents and split in chunks
-    """
-    print(f"Loading documents from {source_directory}")
-    documents = load_documents(source_directory, ignored_files)
-    if not documents:
-        print("No new documents to load")
-        exit(0)
-    print(f"Loaded {len(documents)} new documents from {source_directory}")
-    text_splitter = RecursiveCharacterTextSplitter(chunk_size=chunk_size, chunk_overlap=chunk_overlap)
-    texts = text_splitter.split_documents(documents)
-    print(f"Split into {len(texts)} chunks of text (max. {chunk_size} tokens each)")
-    return texts
-
-def does_vectorstore_exist(persist_directory: str) -> bool:
-    """
-    Checks if vectorstore exists
-    """
-    if os.path.exists(os.path.join(persist_directory, 'index')):
-        if os.path.exists(os.path.join(persist_directory, 'chroma-collections.parquet')) and os.path.exists(os.path.join(persist_directory, 'chroma-embeddings.parquet')):
-            list_index_files = glob.glob(os.path.join(persist_directory, 'index/*.bin'))
-            list_index_files += glob.glob(os.path.join(persist_directory, 'index/*.pkl'))
-            # At least 3 documents are needed in a working vectorstore
-            if len(list_index_files) > 3:
-                return True
-    return False
-
-def main():
-    # Create embeddings
-    embeddings = HuggingFaceEmbeddings(model_name=embeddings_model_name)
-
-    if does_vectorstore_exist(persist_directory):
-        # Update and store locally vectorstore
-        print(f"Appending to existing vectorstore at {persist_directory}")
-        db = Chroma(persist_directory=persist_directory, embedding_function=embeddings, client_settings=CHROMA_SETTINGS)
-        collection = db.get()
-        texts = process_documents([metadata['source'] for metadata in collection['metadatas']])
-        print(f"Creating embeddings. May take some minutes...")
-        db.add_documents(texts)
-    else:
-        # Create and store locally vectorstore
-        print("Creating new vectorstore")
-        texts = process_documents()
-        print(f"Creating embeddings. May take some minutes...")
-        db = Chroma.from_documents(texts, embeddings, persist_directory=persist_directory, client_settings=CHROMA_SETTINGS)
-    db.persist()
-    db = None
-
-    print(f"Ingestion complete! You can now run privateGPT.py to query your documents")
-
-
-if __name__ == "__main__":
-    main()

examples/privategpt/privateGPT.py

@@ -1,71 +0,0 @@
-#!/usr/bin/env python3
-from langchain.chains import RetrievalQA
-from langchain.embeddings import HuggingFaceEmbeddings
-from langchain.callbacks.streaming_stdout import StreamingStdOutCallbackHandler
-from langchain.vectorstores import Chroma
-from langchain.llms import Ollama
-import os
-import argparse
-import time
-
-model = os.environ.get("MODEL", "llama2-uncensored")
-# For embeddings model, the example uses a sentence-transformers model
-# https://www.sbert.net/docs/pretrained_models.html 
-# "The all-mpnet-base-v2 model provides the best quality, while all-MiniLM-L6-v2 is 5 times faster and still offers good quality."
-embeddings_model_name = os.environ.get("EMBEDDINGS_MODEL_NAME", "all-MiniLM-L6-v2")
-persist_directory = os.environ.get("PERSIST_DIRECTORY", "db")
-target_source_chunks = int(os.environ.get('TARGET_SOURCE_CHUNKS',4))
-
-from constants import CHROMA_SETTINGS
-
-def main():
-    # Parse the command line arguments
-    args = parse_arguments()
-    embeddings = HuggingFaceEmbeddings(model_name=embeddings_model_name)
-    db = Chroma(persist_directory=persist_directory, embedding_function=embeddings, client_settings=CHROMA_SETTINGS)
-    retriever = db.as_retriever(search_kwargs={"k": target_source_chunks})
-    # activate/deactivate the streaming StdOut callback for LLMs
-    callbacks = [] if args.mute_stream else [StreamingStdOutCallbackHandler()]
-
-    llm = Ollama(model=model, callbacks=callbacks)
-
-    qa = RetrievalQA.from_chain_type(llm=llm, chain_type="stuff", retriever=retriever, return_source_documents= not args.hide_source)
-    # Interactive questions and answers
-    while True:
-        query = input("\nEnter a query: ")
-        if query == "exit":
-            break
-        if query.strip() == "":
-            continue
-
-        # Get the answer from the chain
-        start = time.time()
-        res = qa(query)
-        answer, docs = res['result'], [] if args.hide_source else res['source_documents']
-        end = time.time()
-
-        # Print the result
-        print("\n\n> Question:")
-        print(query)
-        print(answer)
-
-        # Print the relevant sources used for the answer
-        for document in docs:
-            print("\n> " + document.metadata["source"] + ":")
-            print(document.page_content)
-
-def parse_arguments():
-    parser = argparse.ArgumentParser(description='privateGPT: Ask questions to your documents without an internet connection, '
-                                                 'using the power of LLMs.')
-    parser.add_argument("--hide-source", "-S", action='store_true',
-                        help='Use this flag to disable printing of source documents used for answers.')
-
-    parser.add_argument("--mute-stream", "-M",
-                        action='store_true',
-                        help='Use this flag to disable the streaming StdOut callback for LLMs.')
-
-    return parser.parse_args()
-
-
-if __name__ == "__main__":
-    main()

examples/privategpt/pyproject.toml

@@ -1,26 +0,0 @@
-[tool.poetry]
-name = "privategpt"
-version = "0.1.0"
-description = ""
-authors = ["Ivan Martinez <ivanmartit@gmail.com>"]
-license = "Apache Version 2.0"
-readme = "README.md"
-
-[tool.poetry.dependencies]
-python = "^3.10"
-langchain = "0.0.261"
-gpt4all = "^1.0.3"
-chromadb = "^0.3.26"
-PyMuPDF = "^1.22.5"
-python-dotenv = "^1.0.0"
-unstructured = "^0.8.0"
-extract-msg = "^0.41.5"
-tabulate = "^0.9.0"
-pandoc = "^2.3"
-pypandoc = "^1.11"
-tqdm = "^4.65.0"
-sentence-transformers = "^2.2.2"
-
-[build-system]
-requires = ["poetry-core"]
-build-backend = "poetry.core.masonry.api"

examples/python/client.py

@@ -1,38 +0,0 @@
-import json
-import requests
-
-# NOTE: ollama must be running for this to work, start the ollama app or run `ollama serve`
-model = 'llama2' # TODO: update this for whatever model you wish to use
-
-def generate(prompt, context):
-    r = requests.post('http://localhost:11434/api/generate',
-                      json={
-                          'model': model,
-                          'prompt': prompt,
-                          'context': context,
-                      },
-                      stream=True)
-    r.raise_for_status()
-
-    for line in r.iter_lines():
-        body = json.loads(line)
-        response_part = body.get('response', '')
-        # the response streams one token at a time, print that as we recieve it
-        print(response_part, end='', flush=True)
-
-        if 'error' in body:
-            raise Exception(body['error'])
-
-        if body.get('done', False):
-            return body['context']
-
-def main():
-    context = [] # the context stores a conversation history, you can use this to make the model more context aware
-    while True:
-        user_input = input("Enter a prompt: ")
-        print()
-        context = generate(user_input, context)
-        print()
-
-if __name__ == "__main__":
-    main()

examples/sentiments/Modelfile

@@ -1,28 +0,0 @@
-# Modelfile for creating a sentiment analyzer. 
-# Run `ollama create sentiments -f pathtofile` and then `ollama run sentiments` and enter a topic
-
-FROM orca
-TEMPLATE """
-{{- if .First }}
-### System:
-{{ .System }}
-{{- end }}
-### User: 
-I hate it when my phone dies
-### Response: 
-NEGATIVE
-### User: 
-He is awesome
-### Response: 
-POSITIVE
-### User: 
-This is the link to the article
-### Response: 
-NEUTRAL
-### User:
-{{ .Prompt }}
-
-### Response:
-"""
-
-SYSTEM """You are a sentiment analyzer. You will receive text and output only one word, either POSITIVE or NEGATIVE or NEUTRAL, depending on the sentiment of the text."""

examples/sentiments/Readme.md

@@ -1,25 +0,0 @@
-# Sentiments Modelfile
-
-This is a simple sentiments analyzer using the Orca model. When you pull Orca from the registry, it has a Template already defined that looks like this:
-
-```Modelfile
-{{- if .First }}
-### System:
-{{ .System }}
-{{- end }}
-
-### User:
-{{ .Prompt }}
-
-### Response:
-```
-
-If we just wanted to have the text:
-
-```Plaintext
-You are a sentiment analyzer. You will receive text and output only one word, either POSITIVE or NEGATIVE or NEUTRAL, depending on the sentiment of the text.
-```
-
-then we could have put this in a SYSTEM block. But we want to provide examples which require updating the full Template. Any Modelfile you create will inherit all the settings from the source model. But in this example, we are overriding the Template.
-
-When providing examples for the input and output, you should include the way the model usually provides information. Since the Orca model expects a user prompt to appear after ### User: and the response is after ### Response, we should format our examples like that as well. If we were using the Llama 2 model, the format would be a bit different.

examples/tweetwriter/Modelfile

@@ -3,5 +3,5 @@
 
 FROM nous-hermes
 SYSTEM """
-You are a content marketer who needs to come up with a short but succinct tweet. Make sure to include the appropriate hashtags and links. Sometimes when appropriate, describe a meme that can be included as well. All answers should be in the form of a tweet which has a max size of 280 characters. Every instruction will be the topic to create a tweet about.
+You are a content marketer who needs to come up with a short but succinct tweet. Make sure to include the appropriate hashtags and links. Sometimes when appropriate, describe a meme that can be includes as well. All answers should be in the form of a tweet which has a max size of 280 characters. Every instruction will be the topic to create a tweet about.
 """

format/openssh.go

@@ -1,102 +0,0 @@
-// Copyright 2012 The Go Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE file.
-
-// Code originally from https://go-review.googlesource.com/c/crypto/+/218620
-
-// TODO: replace with upstream once the above change is merged and released.
-
-package format
-
-import (
-	"crypto"
-	"crypto/ed25519"
-	"crypto/rand"
-	"encoding/binary"
-	"encoding/pem"
-	"fmt"
-
-	"golang.org/x/crypto/ssh"
-)
-
-const privateKeyAuthMagic = "openssh-key-v1\x00"
-
-type openSSHEncryptedPrivateKey struct {
-	CipherName string
-	KDFName    string
-	KDFOptions string
-	KeysCount  uint32
-	PubKey     []byte
-	KeyBlocks  []byte
-}
-
-type openSSHPrivateKey struct {
-	Check1  uint32
-	Check2  uint32
-	Keytype string
-	Rest    []byte `ssh:"rest"`
-}
-
-type openSSHEd25519PrivateKey struct {
-	Pub     []byte
-	Priv    []byte
-	Comment string
-	Pad     []byte `ssh:"rest"`
-}
-
-func OpenSSHPrivateKey(key crypto.PrivateKey, comment string) (*pem.Block, error) {
-	var check uint32
-	if err := binary.Read(rand.Reader, binary.BigEndian, &check); err != nil {
-		return nil, err
-	}
-
-	var pk1 openSSHPrivateKey
-	pk1.Check1 = check
-	pk1.Check2 = check
-
-	var w openSSHEncryptedPrivateKey
-	w.KeysCount = 1
-
-	if k, ok := key.(*ed25519.PrivateKey); ok {
-		key = *k
-	}
-
-	switch k := key.(type) {
-	case ed25519.PrivateKey:
-		pub, priv := k[32:], k
-		key := openSSHEd25519PrivateKey{
-			Pub:     pub,
-			Priv:    priv,
-			Comment: comment,
-		}
-
-		pk1.Keytype = ssh.KeyAlgoED25519
-		pk1.Rest = ssh.Marshal(key)
-
-		w.PubKey = ssh.Marshal(struct {
-			KeyType string
-			Pub     []byte
-		}{
-			ssh.KeyAlgoED25519, pub,
-		})
-	default:
-		return nil, fmt.Errorf("ssh: unknown key type %T", k)
-	}
-
-	w.KeyBlocks = openSSHPadding(ssh.Marshal(pk1), 8)
-
-	w.CipherName, w.KDFName, w.KDFOptions = "none", "none", ""
-
-	return &pem.Block{
-		Type:  "OPENSSH PRIVATE KEY",
-		Bytes: append([]byte(privateKeyAuthMagic), ssh.Marshal(w)...),
-	}, nil
-}
-
-func openSSHPadding(block []byte, blocksize int) []byte {
-	for i, j := 0, len(block); (j+i)%blocksize != 0; i++ {
-		block = append(block, byte(i+1))
-	}
-
-	return block
-}

ggml-metal.metal

@@ -0,0 +1 @@
+llama/ggml-metal.metal

go.mod

@@ -8,17 +8,17 @@ require (
 	github.com/mattn/go-runewidth v0.0.14
 	github.com/mitchellh/colorstring v0.0.0-20190213212951-d06e56a500db
 	github.com/olekukonko/tablewriter v0.0.5
-	github.com/pdevine/readline v1.5.2
 	github.com/spf13/cobra v1.7.0
 )
 
 require github.com/rivo/uniseg v0.2.0 // indirect
 
 require (
+	dario.cat/mergo v1.0.0
 	github.com/bytedance/sonic v1.9.1 // indirect
 	github.com/chenzhuoyu/base64x v0.0.0-20221115062448-fe3a3abad311 // indirect
+	github.com/chzyer/readline v1.5.1
 	github.com/gabriel-vasile/mimetype v1.4.2 // indirect
-	github.com/gin-contrib/cors v1.4.0
 	github.com/gin-contrib/sse v0.1.0 // indirect
 	github.com/go-playground/locales v0.14.1 // indirect
 	github.com/go-playground/universal-translator v0.18.1 // indirect
@@ -32,19 +32,16 @@ require (
 	github.com/mattn/go-isatty v0.0.19 // indirect
 	github.com/modern-go/concurrent v0.0.0-20180306012644-bacd9c7ef1dd // indirect
 	github.com/modern-go/reflect2 v1.0.2 // indirect
-	github.com/pbnjay/memory v0.0.0-20210728143218-7b4eea64cf58
 	github.com/pelletier/go-toml/v2 v2.0.8 // indirect
 	github.com/spf13/pflag v1.0.5 // indirect
 	github.com/twitchyliquid64/golang-asm v0.15.1 // indirect
 	github.com/ugorji/go/codec v1.2.11 // indirect
 	golang.org/x/arch v0.3.0 // indirect
-	golang.org/x/crypto v0.10.0
-	golang.org/x/exp v0.0.0-20230817173708-d852ddb80c63
+	golang.org/x/crypto v0.10.0 // indirect
 	golang.org/x/net v0.10.0 // indirect
-	golang.org/x/sys v0.11.0 // indirect
+	golang.org/x/sys v0.10.0 // 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

@@ -1,3 +1,5 @@
+dario.cat/mergo v1.0.0 h1:AGCNq9Evsj31mOgNPcLyXc+4PNABt905YmuqPYYpBWk=
+dario.cat/mergo v1.0.0/go.mod h1:uNxQE+84aUszobStD9th8a29P2fMDhsBdgRYvZOxGmk=
 github.com/bytedance/sonic v1.5.0/go.mod h1:ED5hyg4y6t3/9Ku1R6dU/4KyJ48DZ4jPhfY1O2AihPM=
 github.com/bytedance/sonic v1.9.1 h1:6iJ6NqdoxCDr6mbY8h18oSO+cShGSMRGCEo7F2h0x8s=
 github.com/bytedance/sonic v1.9.1/go.mod h1:i736AoUSYt75HyZLoJW9ERYxcy6eaN6h4BZXU064P/U=
@@ -6,10 +8,11 @@ github.com/chenzhuoyu/base64x v0.0.0-20221115062448-fe3a3abad311 h1:qSGYFH7+jGhD
 github.com/chenzhuoyu/base64x v0.0.0-20221115062448-fe3a3abad311/go.mod h1:b583jCggY9gE99b6G5LEC39OIiVsWj+R97kbl5odCEk=
 github.com/chzyer/logex v1.2.1 h1:XHDu3E6q+gdHgsdTPH6ImJMIp436vR6MPtH8gP05QzM=
 github.com/chzyer/logex v1.2.1/go.mod h1:JLbx6lG2kDbNRFnfkgvh4eRJRPX1QCoOIWomwysCBrQ=
+github.com/chzyer/readline v1.5.1 h1:upd/6fQk4src78LMRzh5vItIt361/o4uq553V8B5sGI=
+github.com/chzyer/readline v1.5.1/go.mod h1:Eh+b79XXUwfKfcPLepksvw2tcLE/Ct21YObkaSkeBlk=
 github.com/chzyer/test v1.0.0 h1:p3BQDXSxOhOG0P9z6/hGnII4LGiEPOYBhs8asl/fC04=
 github.com/chzyer/test v1.0.0/go.mod h1:2JlltgoNkt4TW/z9V/IzDdFaMTM2JPIi26O1pF38GC8=
 github.com/cpuguy83/go-md2man/v2 v2.0.2/go.mod h1:tgQtvFlXSQOSOSIRvRPT7W67SCa46tRHOmNcaadrF8o=
-github.com/creack/pty v1.1.9/go.mod h1:oKZEueFk5CKHvIhNR5MUki03XCEU+Q6VDXinZuGJ33E=
 github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
@@ -17,25 +20,17 @@ github.com/dustin/go-humanize v1.0.1 h1:GzkhY7T5VNhEkwH0PVJgjz+fX1rhBrR7pRT3mDkp
 github.com/dustin/go-humanize v1.0.1/go.mod h1:Mu1zIs6XwVuF/gI1OepvI0qD18qycQx+mFykh5fBlto=
 github.com/gabriel-vasile/mimetype v1.4.2 h1:w5qFW6JKBz9Y393Y4q372O9A7cUSequkh1Q7OhCmWKU=
 github.com/gabriel-vasile/mimetype v1.4.2/go.mod h1:zApsH/mKG4w07erKIaJPFiX0Tsq9BFQgN3qGY5GnNgA=
-github.com/gin-contrib/cors v1.4.0 h1:oJ6gwtUl3lqV0WEIwM/LxPF1QZ5qe2lGWdY2+bz7y0g=
-github.com/gin-contrib/cors v1.4.0/go.mod h1:bs9pNM0x/UsmHPBWT2xZz9ROh8xYjYkiURUfmBoMlcs=
 github.com/gin-contrib/sse v0.1.0 h1:Y/yl/+YNO8GZSjAhjMsSuLt29uWRFHdHYUb5lYOV9qE=
 github.com/gin-contrib/sse v0.1.0/go.mod h1:RHrZQHXnP2xjPF+u1gW/2HnVO7nvIa9PG3Gm+fLHvGI=
-github.com/gin-gonic/gin v1.8.1/go.mod h1:ji8BvRH1azfM+SYow9zQ6SZMvR8qOMZHmsCuWR9tTTk=
 github.com/gin-gonic/gin v1.9.1 h1:4idEAncQnU5cB7BeOkPtxjfCSye0AAm1R0RVIqJ+Jmg=
 github.com/gin-gonic/gin v1.9.1/go.mod h1:hPrL7YrpYKXt5YId3A/Tnip5kqbEAP+KLuI3SUcPTeU=
-github.com/go-playground/assert/v2 v2.0.1/go.mod h1:VDjEfimB/XKnb+ZQfWdccd7VUvScMdVu0Titje2rxJ4=
 github.com/go-playground/assert/v2 v2.2.0 h1:JvknZsQTYeFEAhQwI4qEt9cyV5ONwRHC+lYKSsYSR8s=
-github.com/go-playground/locales v0.14.0/go.mod h1:sawfccIbzZTqEDETgFXqTho0QybSa7l++s0DH+LDiLs=
 github.com/go-playground/locales v0.14.1 h1:EWaQ/wswjilfKLTECiXz7Rh+3BjFhfDFKv/oXslEjJA=
 github.com/go-playground/locales v0.14.1/go.mod h1:hxrqLVvrK65+Rwrd5Fc6F2O76J/NuW9t0sjnWqG1slY=
-github.com/go-playground/universal-translator v0.18.0/go.mod h1:UvRDBj+xPUEGrFYl+lu/H90nyDXpg0fqeB/AQUGNTVA=
 github.com/go-playground/universal-translator v0.18.1 h1:Bcnm0ZwsGyWbCzImXv+pAJnYK9S473LQFuzCbDbfSFY=
 github.com/go-playground/universal-translator v0.18.1/go.mod h1:xekY+UJKNuX9WP91TpwSH2VMlDf28Uj24BCp08ZFTUY=
-github.com/go-playground/validator/v10 v10.10.0/go.mod h1:74x4gJWsvQexRdW8Pn3dXSGrTK4nAUsbPlLADvpJkos=
 github.com/go-playground/validator/v10 v10.14.0 h1:vgvQWe3XCz3gIeFDm/HnTIbj6UGmg/+t63MyGU2n5js=
 github.com/go-playground/validator/v10 v10.14.0/go.mod h1:9iXMNT7sEkjXb0I+enO7QXmzG6QCsPWY4zveKFVRSyU=
-github.com/goccy/go-json v0.9.7/go.mod h1:6MelG93GURQebXPDq3khkgXZkazVtN9CRI+MGFi0w8I=
 github.com/goccy/go-json v0.10.2 h1:CrxCmQqYDkv1z7lO7Wbh2HN93uovUHgrECaO5ZrCXAU=
 github.com/goccy/go-json v0.10.2/go.mod h1:6MelG93GURQebXPDq3khkgXZkazVtN9CRI+MGFi0w8I=
 github.com/golang/protobuf v1.5.0/go.mod h1:FsONVRAS9T7sI+LIUmWTfcYkHO4aIWwzhcaSAoJOfIk=
@@ -50,18 +45,8 @@ github.com/json-iterator/go v1.1.12/go.mod h1:e30LSqwooZae/UwlEbR2852Gd8hjQvJoHm
 github.com/klauspost/cpuid/v2 v2.0.9/go.mod h1:FInQzS24/EEf25PyTYn52gqo7WaD8xa0213Md/qVLRg=
 github.com/klauspost/cpuid/v2 v2.2.4 h1:acbojRNwl3o09bUq+yDCtZFc1aiwaAAxtcn8YkZXnvk=
 github.com/klauspost/cpuid/v2 v2.2.4/go.mod h1:RVVoqg1df56z8g3pUjL/3lE5UfnlrJX8tyFgg4nqhuY=
-github.com/kr/pretty v0.1.0/go.mod h1:dAy3ld7l9f0ibDNOQOHHMYYIIbhfbHSm3C4ZsoJORNo=
-github.com/kr/pretty v0.2.1/go.mod h1:ipq/a2n7PKx3OHsz4KJII5eveXtPO4qwEXGdVfWzfnI=
-github.com/kr/pretty v0.3.0 h1:WgNl7dwNpEZ6jJ9k1snq4pZsg7DOEN8hP9Xw0Tsjwk0=
-github.com/kr/pretty v0.3.0/go.mod h1:640gp4NfQd8pI5XOwp5fnNeVWj67G7CFk/SaSQn7NBk=
-github.com/kr/pty v1.1.1/go.mod h1:pFQYn66WHrOpPYNljwOMqo10TkYh1fy3cYio2l3bCsQ=
-github.com/kr/text v0.1.0/go.mod h1:4Jbv+DJW3UT/LiOwJeYQe1efqtUx/iVham/4vfdArNI=
-github.com/kr/text v0.2.0 h1:5Nx0Ya0ZqY2ygV366QzturHI13Jq95ApcVaJBhpS+AY=
-github.com/kr/text v0.2.0/go.mod h1:eLer722TekiGuMkidMxC/pM04lWEeraHUUmBw8l2grE=
-github.com/leodido/go-urn v1.2.1/go.mod h1:zt4jvISO2HfUBqxjfIshjdMTYS56ZS/qv49ictyFfxY=
 github.com/leodido/go-urn v1.2.4 h1:XlAE/cm/ms7TE/VMVoduSpNBoyc2dOxHs5MZSwAN63Q=
 github.com/leodido/go-urn v1.2.4/go.mod h1:7ZrI8mTSeBSHl/UaRyKQW1qZeMgak41ANeCNaVckg+4=
-github.com/mattn/go-isatty v0.0.14/go.mod h1:7GGIvUiUoEMVVmxf/4nioHXj79iQHKdU27kJ6hsGG94=
 github.com/mattn/go-isatty v0.0.19 h1:JITubQf0MOLdlGRuRq+jtsDlekdYPia9ZFsB8h/APPA=
 github.com/mattn/go-isatty v0.0.19/go.mod h1:W+V8PltTTMOvKvAeJH7IuucS94S2C6jfK/D7dTCTo3Y=
 github.com/mattn/go-runewidth v0.0.9/go.mod h1:H031xJmbD/WCDINGzjvQ9THkh0rPKHF+m2gUSrubnMI=
@@ -76,21 +61,12 @@ github.com/modern-go/reflect2 v1.0.2 h1:xBagoLtFs94CBntxluKeaWgTMpvLxC4ur3nMaC9G
 github.com/modern-go/reflect2 v1.0.2/go.mod h1:yWuevngMOJpCy52FWWMvUC8ws7m/LJsjYzDa0/r8luk=
 github.com/olekukonko/tablewriter v0.0.5 h1:P2Ga83D34wi1o9J6Wh1mRuqd4mF/x/lgBS7N7AbDhec=
 github.com/olekukonko/tablewriter v0.0.5/go.mod h1:hPp6KlRPjbx+hW8ykQs1w3UBbZlj6HuIJcUGPhkA7kY=
-github.com/pbnjay/memory v0.0.0-20210728143218-7b4eea64cf58 h1:onHthvaw9LFnH4t2DcNVpwGmV9E1BkGknEliJkfwQj0=
-github.com/pbnjay/memory v0.0.0-20210728143218-7b4eea64cf58/go.mod h1:DXv8WO4yhMYhSNPKjeNKa5WY9YCIEBRbNzFFPJbWO6Y=
-github.com/pdevine/readline v1.5.2 h1:oz6Y5GdTmhPG+08hhxcAvtHitSANWuA2100Sppb38xI=
-github.com/pdevine/readline v1.5.2/go.mod h1:na/LbuE5PYwxI7GyopWdIs3U8HVe89lYlNTFTXH3wOw=
-github.com/pelletier/go-toml/v2 v2.0.1/go.mod h1:r9LEWfGN8R5k0VXJ+0BkIe7MYkRdwZOjgMj2KwnJFUo=
 github.com/pelletier/go-toml/v2 v2.0.8 h1:0ctb6s9mE31h0/lhu+J6OPmVeDxJn+kYnJc2jZR9tGQ=
 github.com/pelletier/go-toml/v2 v2.0.8/go.mod h1:vuYfssBdrU2XDZ9bYydBu6t+6a6PYNcZljzZR9VXg+4=
-github.com/pkg/diff v0.0.0-20210226163009-20ebb0f2a09e/go.mod h1:pJLUxLENpZxwdsKMEsNbx1VGcRFpLqf3715MtcvvzbA=
 github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=
 github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
 github.com/rivo/uniseg v0.2.0 h1:S1pD9weZBuJdFmowNwbpi7BJ8TNftyUImj/0WQi72jY=
 github.com/rivo/uniseg v0.2.0/go.mod h1:J6wj4VEh+S6ZtnVlnTBMWIodfgj8LQOQFoIToxlJtxc=
-github.com/rogpeppe/go-internal v1.6.1/go.mod h1:xXDCJY+GAPziupqXw64V24skbSoqbTEfhy4qGm1nDQc=
-github.com/rogpeppe/go-internal v1.8.0 h1:FCbCCtXNOY3UtUuHUYaghJg4y7Fd14rXifAYUAtL9R8=
-github.com/rogpeppe/go-internal v1.8.0/go.mod h1:WmiCO8CzOY8rg0OYDC4/i/2WRWAB6poM+XZ2dLUbcbE=
 github.com/russross/blackfriday/v2 v2.1.0/go.mod h1:+Rmxgy9KzJVeS9/2gXHxylqXiyQDYRxCVz55jmeOWTM=
 github.com/spf13/cobra v1.7.0 h1:hyqWnYt1ZQShIddO5kBpj3vu05/++x6tJ6dg8EC572I=
 github.com/spf13/cobra v1.7.0/go.mod h1:uLxZILRyS/50WlhOIKD7W6V5bgeIt+4sICxh6uRMrb0=
@@ -100,7 +76,6 @@ github.com/stretchr/objx v0.1.0/go.mod h1:HFkY916IF+rwdDfMAkV7OtwuqBVzrE8GR6GFx+
 github.com/stretchr/objx v0.4.0/go.mod h1:YvHI0jy2hoMjB+UWwv71VJQ9isScKT/TqJzVSSt89Yw=
 github.com/stretchr/objx v0.5.0/go.mod h1:Yh+to48EsGEfYuaHDzXPcE3xhTkx73EhmCGUpEOglKo=
 github.com/stretchr/testify v1.3.0/go.mod h1:M5WIy9Dh21IEIfnGCwXGc5bZfKNJtfHm1UVUgZn+9EI=
-github.com/stretchr/testify v1.6.1/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
 github.com/stretchr/testify v1.7.0/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
 github.com/stretchr/testify v1.7.1/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
 github.com/stretchr/testify v1.8.0/go.mod h1:yNjHg4UonilssWZ8iaSj1OCr/vHnekPRkoO+kdMU+MU=
@@ -110,53 +85,31 @@ github.com/stretchr/testify v1.8.3 h1:RP3t2pwF7cMEbC1dqtB6poj3niw/9gnV4Cjg5oW5gt
 github.com/stretchr/testify v1.8.3/go.mod h1:sz/lmYIOXD/1dqDmKjjqLyZ2RngseejIcXlSw2iwfAo=
 github.com/twitchyliquid64/golang-asm v0.15.1 h1:SU5vSMR7hnwNxj24w34ZyCi/FmDZTkS4MhqMhdFk5YI=
 github.com/twitchyliquid64/golang-asm v0.15.1/go.mod h1:a1lVb/DtPvCB8fslRZhAngC2+aY1QWCk3Cedj/Gdt08=
-github.com/ugorji/go v1.2.7/go.mod h1:nF9osbDWLy6bDVv/Rtoh6QgnvNDpmCalQV5urGCCS6M=
-github.com/ugorji/go/codec v1.2.7/go.mod h1:WGN1fab3R1fzQlVQTkfxVtIBhWDRqOviHU95kRgeqEY=
 github.com/ugorji/go/codec v1.2.11 h1:BMaWp1Bb6fHwEtbplGBGJ498wD+LKlNSl25MjdZY4dU=
 github.com/ugorji/go/codec v1.2.11/go.mod h1:UNopzCgEMSXjBc6AOMqYvWC1ktqTAfzJZUZgYf6w6lg=
 golang.org/x/arch v0.0.0-20210923205945-b76863e36670/go.mod h1:5om86z9Hs0C8fWVUuoMHwpExlXzs5Tkyp9hOrfG7pp8=
 golang.org/x/arch v0.3.0 h1:02VY4/ZcO/gBOH6PUaoiptASxtXU10jazRCP865E97k=
 golang.org/x/arch v0.3.0/go.mod h1:5om86z9Hs0C8fWVUuoMHwpExlXzs5Tkyp9hOrfG7pp8=
-golang.org/x/crypto v0.0.0-20210711020723-a769d52b0f97/go.mod h1:GvvjBRRGRdwPK5ydBHafDWAxML/pGHZbMvKqRZ5+Abc=
 golang.org/x/crypto v0.10.0 h1:LKqV2xt9+kDzSTfOhx4FrkEBcMrAgHSYgzywV9zcGmM=
 golang.org/x/crypto v0.10.0/go.mod h1:o4eNf7Ede1fv+hwOwZsTHl9EsPFO6q6ZvYR8vYfY45I=
-golang.org/x/exp v0.0.0-20230817173708-d852ddb80c63 h1:m64FZMko/V45gv0bNmrNYoDEq8U5YUhetc9cBWKS1TQ=
-golang.org/x/exp v0.0.0-20230817173708-d852ddb80c63/go.mod h1:0v4NqG35kSWCMzLaMeX+IQrlSnVE/bqGSyC2cz/9Le8=
-golang.org/x/net v0.0.0-20210226172049-e18ecbb05110/go.mod h1:m0MpNAwzfU5UDzcl9v0D8zg8gWTRqZa9RBIspLL5mdg=
 golang.org/x/net v0.10.0 h1:X2//UzNDwYmtCLn7To6G58Wr6f5ahEAQgKNzv9Y951M=
 golang.org/x/net v0.10.0/go.mod h1:0qNGK6F8kojg2nk9dLZ2mShWaEBan6FAoqfSigmmuDg=
-golang.org/x/sys v0.0.0-20201119102817-f84b799fce68/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20210615035016-665e8c7367d1/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.0.0-20210630005230-0f9fa26af87c/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.0.0-20210806184541-e5e7981a1069/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.0.0-20220310020820-b874c991c1a5/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.0.0-20220704084225-05e143d24a9e/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.6.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.11.0 h1:eG7RXZHdqOJ1i+0lgLgCpSXAp6M3LYlAo6osgSi0xOM=
-golang.org/x/sys v0.11.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/term v0.0.0-20201126162022-7de9c90e9dd1/go.mod h1:bj7SfCRtBDWHUb9snDiAeCFNEtKQo2Wmx5Cou7ajbmo=
+golang.org/x/sys v0.10.0 h1:SqMFp9UcQJZa+pmYuAKjd9xq1f0j5rLcDIk0mj4qAsA=
+golang.org/x/sys v0.10.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/term v0.10.0 h1:3R7pNqamzBraeqj/Tj8qt1aQ2HpmlC+Cx/qL/7hn4/c=
 golang.org/x/term v0.10.0/go.mod h1:lpqdcUyK/oCiQxvxVrppt5ggO2KCZ5QblwqPnfZ6d5o=
-golang.org/x/text v0.3.3/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
-golang.org/x/text v0.3.6/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
 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=
 google.golang.org/protobuf v1.30.0/go.mod h1:HV8QOd/L58Z+nl8r43ehVNZIU/HEI6OcFqwMG9pJV4I=
+gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405 h1:yhCVgyC4o1eVCa2tZl7eS0r+SDo693bJlVdllGtEeKM=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
-gopkg.in/check.v1 v1.0.0-20180628173108-788fd7840127/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
-gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c h1:Hei/4ADfdWqJk1ZMxUNpqntNwaWcugrBjAiHlqqRiVk=
-gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c/go.mod h1:JHkPIbrfpd72SG/EVd6muEfDQjcINNoR0C8j2r3qZ4Q=
-gopkg.in/errgo.v2 v2.1.0/go.mod h1:hNsd1EY+bozCKY1Ytp96fpM3vjJbqLJn88ws8XvfDNI=
-gopkg.in/yaml.v2 v2.4.0/go.mod h1:RDklbk79AGWmwhnvt/jBztapEOGDOx6ZbXqjP6csGnQ=
 gopkg.in/yaml.v3 v3.0.0-20200313102051-9f266ea9e77c/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
-gopkg.in/yaml.v3 v3.0.0-20210107192922-496545a6307b/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
 gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=
 gopkg.in/yaml.v3 v3.0.1/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
 rsc.io/pdf v0.1.1/go.mod h1:n8OzWcQ6Sp37PL01nO98y4iUCRdTGarVfzxY20ICaU4=

library/.gitignore

@@ -0,0 +1 @@
+models

library/downloads

@@ -0,0 +1,7 @@
+https://huggingface.co/TheBloke/orca_mini_3B-GGML/resolve/main/orca-mini-3b.ggmlv3.q4_0.bin e84705205f71dd55be7b24a778f248f0eda9999a125d313358c087e092d83148
+https://huggingface.co/TheBloke/Nous-Hermes-13B-GGML/resolve/main/nous-hermes-13b.ggmlv3.q4_0.bin d1735b93e1dc503f1045ccd6c8bd73277b18ba892befd1dc29e9b9a7822ed998
+https://huggingface.co/TheBloke/vicuna-7B-v1.3-GGML/resolve/main/vicuna-7b-v1.3.ggmlv3.q4_0.bin 23ce5ed290b56a19305178b9ada2c3d96036bd69a6c18304b6158eb6672d6c0f
+https://huggingface.co/TheBloke/Wizard-Vicuna-13B-Uncensored-GGML/resolve/main/Wizard-Vicuna-13B-Uncensored.ggmlv3.q4_0.bin 1f08b147a5bce41cfcbb3fd5d51ba765dea1786e15b5655ab69ba3a337a893b7
+https://huggingface.co/TheBloke/Llama-2-7B-GGML/resolve/main/llama-2-7b.ggmlv3.q4_0.bin bfa26d855e44629c4cf919985e90bd7fa03b77eea1676791519e39a4d45fd4d5
+https://huggingface.co/TheBloke/Llama-2-7B-Chat-GGML/resolve/main/llama-2-7b-chat.ggmlv3.q4_0.bin 8daa9615cce30c259a9555b1cc250d461d1bc69980a274b44d7eda0be78076d8
+https://huggingface.co/TheBloke/Llama-2-13B-chat-GGML/resolve/main/llama-2-13b-chat.ggmlv3.q4_0.bin f79142715bc9539a2edbb4b253548db8b34fac22736593eeaa28555874476e30

library/modelfiles/llama2

@@ -0,0 +1,147 @@
+FROM ../models/llama-2-7b-chat.ggmlv3.q4_0.bin
+
+TEMPLATE """
+{{- if .First }}
+<<SYS>>
+{{ .System }}
+<</SYS>>
+{{- end }}
+
+[INST] {{ .Prompt }} [/INST]
+"""
+
+SYSTEM """
+You are a helpful, respectful and honest assistant. Always answer as helpfully as possible, while being safe. Your answers should not include any harmful, unethical, racist, sexist, toxic, dangerous, or illegal content. Please ensure that your responses are socially unbiased and positive in nature.
+
+If a question does not make any sense, or is not factually coherent, explain why instead of answering something not correct. If you don't know the answer to a question, please don't share false information.
+"""
+
+LICENSE """
+Llama 2 Community License Agreement
+
+Llama 2 Version Release Date: July 18, 2023
+
+“Agreement” means the terms and conditions for use, reproduction, distribution and modification of the Llama Materials set forth herein.
+
+“Documentation” means the specifications, manuals and documentation accompanying Llama 2 distributed by Meta at ai.meta.com/resources/models-and-libraries/llama-downloads/.
+
+“Licensee” or “you” means you, or your employer or any other person or entity (if you are entering into this Agreement on such person or entity’s behalf), of the age required under applicable laws, rules or regulations to provide legal consent and that has legal authority to bind your employer or such other person or entity if you are entering in this Agreement on their behalf.
+
+“Llama 2” means the foundational large language models and software and algorithms, including machine-learning model code, trained model weights, inference-enabling code, training-enabling code, fine-tuning enabling code and other elements of the foregoing distributed by Meta at ai.meta.com/resources/models-and-libraries/llama-downloads/.
+
+“Llama Materials” means, collectively, Meta’s proprietary Llama 2 and Documentation (and any portion thereof) made available under this Agreement.
+
+“Meta” or “we” means Meta Platforms Ireland Limited (if you are located in or, if you are an entity, your principal place of business is in the EEA or Switzerland) and Meta Platforms, Inc. (if you are located outside of the EEA or Switzerland).
+
+By clicking “I Accept” below or by using or distributing any portion or element of the Llama Materials, you agree to be bound by this Agreement.
+
+1. License Rights and Redistribution.
+
+a. Grant of Rights. You are granted a non-exclusive, worldwide, non-transferable and royalty-free limited license under Meta’s intellectual property or other rights owned by Meta embodied in the Llama Materials to use, reproduce, distribute, copy, create derivative works of, and make modifications to the Llama Materials.
+
+b. Redistribution and Use.
+
+i. If you distribute or make the Llama Materials, or any derivative works thereof, available to a third party, you shall provide a copy of this Agreement to such third party.
+
+ii. If you receive Llama Materials, or any derivative works thereof, from a Licensee as part of an integrated end user product, then Section 2 of this Agreement will not apply to you.
+
+iii. You must retain in all copies of the Llama Materials that you distribute the following attribution notice within a “Notice” text file distributed as a part of such copies: “Llama 2 is licensed under the LLAMA 2 Community License, Copyright © Meta Platforms, Inc. All Rights Reserved.”
+
+iv. Your use of the Llama Materials must comply with applicable laws and regulations (including trade compliance laws and regulations) and adhere to the Acceptable Use Policy for the Llama Materials (available at https://ai.meta.com/llama/use-policy), which is hereby incorporated by reference into this Agreement.
+
+v. You will not use the Llama Materials or any output or results of the Llama Materials to improve any other large language model (excluding Llama 2 or derivative works thereof).
+
+2. Additional Commercial Terms. If, on the Llama 2 version release date, the monthly active users of the products or services made available by or for Licensee, or Licensee’s affiliates, is greater than 700 million monthly active users in the preceding calendar month, you must request a license from Meta, which Meta may grant to you in its sole discretion, and you are not authorized to exercise any of the rights under this Agreement unless or until Meta otherwise expressly grants you such rights.
+
+3. Disclaimer of Warranty. UNLESS REQUIRED BY APPLICABLE LAW, THE LLAMA MATERIALS AND ANY OUTPUT AND RESULTS THEREFROM ARE PROVIDED ON AN “AS IS” BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED, INCLUDING, WITHOUT LIMITATION, ANY WARRANTIES OF TITLE, NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE. YOU ARE SOLELY RESPONSIBLE FOR DETERMINING THE APPROPRIATENESS OF USING OR REDISTRIBUTING THE LLAMA MATERIALS AND ASSUME ANY RISKS ASSOCIATED WITH YOUR USE OF THE LLAMA MATERIALS AND ANY OUTPUT AND RESULTS.
+
+4. Limitation of Liability. IN NO EVENT WILL META OR ITS AFFILIATES BE LIABLE UNDER ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, TORT, NEGLIGENCE, PRODUCTS LIABILITY, OR OTHERWISE, ARISING OUT OF THIS AGREEMENT, FOR ANY LOST PROFITS OR ANY INDIRECT, SPECIAL, CONSEQUENTIAL, INCIDENTAL, EXEMPLARY OR PUNITIVE DAMAGES, EVEN IF META OR ITS AFFILIATES HAVE BEEN ADVISED OF THE POSSIBILITY OF ANY OF THE FOREGOING.
+
+5. Intellectual Property.
+
+a. No trademark licenses are granted under this Agreement, and in connection with the Llama Materials, neither Meta nor Licensee may use any name or mark owned by or associated with the other or any of its affiliates, except as required for reasonable and customary use in describing and redistributing the Llama Materials.
+
+b. Subject to Meta’s ownership of Llama Materials and derivatives made by or for Meta, with respect to any derivative works and modifications of the Llama Materials that are made by you, as between you and Meta, you are and will be the owner of such derivative works and modifications.
+
+c. If you institute litigation or other proceedings against Meta or any entity (including a cross-claim or counterclaim in a lawsuit) alleging that the Llama Materials or Llama 2 outputs or results, or any portion of any of the foregoing, constitutes infringement of intellectual property or other rights owned or licensable by you, then any licenses granted to you under this Agreement shall terminate as of the date such litigation or claim is filed or instituted. You will indemnify and hold harmless Meta from and against any claim by any third party arising out of or related to your use or distribution of the Llama Materials.
+
+6. Term and Termination. The term of this Agreement will commence upon your acceptance of this Agreement or access to the Llama Materials and will continue in full force and effect until terminated in accordance with the terms and conditions herein. Meta may terminate this Agreement if you are in breach of any term or condition of this Agreement. Upon termination of this Agreement, you shall delete and cease use of the Llama Materials. Sections 3, 4 and 7 shall survive the termination of this Agreement.
+
+7. Governing Law and Jurisdiction. This Agreement will be governed and construed under the laws of the State of California without regard to choice of law principles, and the UN Convention on Contracts for the International Sale of Goods does not apply to this Agreement. The courts of California shall have exclusive jurisdiction of any dispute arising out of this Agreement.
+
+"""
+
+LICENSE """
+Llama 2 Acceptable Use Policy
+
+Meta is committed to promoting safe and fair use of its tools and features, including Llama 2. If you access or use Llama 2, you agree to this Acceptable Use Policy (“Policy”). The most recent copy of this policy can be found at ai.meta.com/llama/use-policy.
+
+Prohibited Uses
+
+We want everyone to use Llama 2 safely and responsibly. You agree you will not use, or allow others to use, Llama 2 to:
+
+1. Violate the law or others’ rights, including to:
+
+a. Engage in, promote, generate, contribute to, encourage, plan, incite, or further illegal or unlawful activity or content, such as:
+
+i. Violence or terrorism
+
+ii. Exploitation or harm to children, including the solicitation, creation, acquisition, or dissemination of child exploitative content or failure to report Child Sexual Abuse Material
+
+b. Human trafficking, exploitation, and sexual violence
+
+iii. The illegal distribution of information or materials to minors, including obscene materials, or failure to employ legally required age-gating in connection with such information or materials.
+
+iv. Sexual solicitation
+
+vi. Any other criminal activity
+
+c. Engage in, promote, incite, or facilitate the harassment, abuse, threatening, or bullying of individuals or groups of individuals
+
+d. Engage in, promote, incite, or facilitate discrimination or other unlawful or harmful conduct in the provision of employment, employment benefits, credit, housing, other economic benefits, or other essential goods and services
+
+e. Engage in the unauthorized or unlicensed practice of any profession including, but not limited to, financial, legal, medical/health, or related professional practices
+
+f. Collect, process, disclose, generate, or infer health, demographic, or other sensitive personal or private information about individuals without rights and consents required by applicable laws
+
+g. Engage in or facilitate any action or generate any content that infringes, misappropriates, or otherwise violates any third-party rights, including the outputs or results of any products or services using the Llama 2 Materials
+
+h. Create, generate, or facilitate the creation of malicious code, malware, computer viruses or do anything else that could disable, overburden, interfere with or impair the proper working, integrity, operation or appearance of a website or computer system
+
+2. Engage in, promote, incite, facilitate, or assist in the planning or development of activities that present a risk of death or bodily harm to individuals, including use of Llama 2 related to the following:
+
+a. Military, warfare, nuclear industries or applications, espionage, use for materials or activities that are subject to the International Traffic Arms Regulations (ITAR) maintained by the United States Department of State
+
+b. Guns and illegal weapons (including weapon development)
+
+c. Illegal drugs and regulated/controlled substances
+
+d. Operation of critical infrastructure, transportation technologies, or heavy machinery
+
+e. Self-harm or harm to others, including suicide, cutting, and eating disorders
+
+f. Any content intended to incite or promote violence, abuse, or any infliction of bodily harm to an individual
+
+3. Intentionally deceive or mislead others, including use of Llama 2 related to the following:
+
+a. Generating, promoting, or furthering fraud or the creation or promotion of disinformation
+
+b. Generating, promoting, or furthering defamatory content, including the creation of defamatory statements, images, or other content
+
+c. Generating, promoting, or further distributing spam
+
+d. Impersonating another individual without consent, authorization, or legal right
+
+e. Representing that the use of Llama 2 or outputs are human-generated
+
+f. Generating or facilitating false online engagement, including fake reviews and other means of fake online engagement
+
+4. Fail to appropriately disclose to end users any known dangers of your AI system
+
+Please report any violation of this Policy, software “bug,” or other problems that could lead to a violation of this Policy through one of the following means:
+
+Reporting issues with the model: github.com/facebookresearch/llama
+Reporting risky content generated by the model: developers.facebook.com/llama_output_feedback
+Reporting bugs and security concerns: facebook.com/whitehat/info
+Reporting violations of the Acceptable Use Policy or unlicensed uses of Llama: LlamaUseReport@meta.com
+"""

library/modelfiles/llama2_13b

@@ -0,0 +1,147 @@
+FROM ../models/llama-2-13b-chat.ggmlv3.q4_0.bin
+
+TEMPLATE """
+{{- if .First }}
+<<SYS>>
+{{ .System }}
+<</SYS>>
+{{- end }}
+
+[INST] {{ .Prompt }} [/INST]
+"""
+
+SYSTEM """
+You are a helpful, respectful and honest assistant. Always answer as helpfully as possible, while being safe. Your answers should not include any harmful, unethical, racist, sexist, toxic, dangerous, or illegal content. Please ensure that your responses are socially unbiased and positive in nature.
+
+If a question does not make any sense, or is not factually coherent, explain why instead of answering something not correct. If you don't know the answer to a question, please don't share false information.
+"""
+
+LICENSE """
+Llama 2 Community License Agreement
+
+Llama 2 Version Release Date: July 18, 2023
+
+“Agreement” means the terms and conditions for use, reproduction, distribution and modification of the Llama Materials set forth herein.
+
+“Documentation” means the specifications, manuals and documentation accompanying Llama 2 distributed by Meta at ai.meta.com/resources/models-and-libraries/llama-downloads/.
+
+“Licensee” or “you” means you, or your employer or any other person or entity (if you are entering into this Agreement on such person or entity’s behalf), of the age required under applicable laws, rules or regulations to provide legal consent and that has legal authority to bind your employer or such other person or entity if you are entering in this Agreement on their behalf.
+
+“Llama 2” means the foundational large language models and software and algorithms, including machine-learning model code, trained model weights, inference-enabling code, training-enabling code, fine-tuning enabling code and other elements of the foregoing distributed by Meta at ai.meta.com/resources/models-and-libraries/llama-downloads/.
+
+“Llama Materials” means, collectively, Meta’s proprietary Llama 2 and Documentation (and any portion thereof) made available under this Agreement.
+
+“Meta” or “we” means Meta Platforms Ireland Limited (if you are located in or, if you are an entity, your principal place of business is in the EEA or Switzerland) and Meta Platforms, Inc. (if you are located outside of the EEA or Switzerland).
+
+By clicking “I Accept” below or by using or distributing any portion or element of the Llama Materials, you agree to be bound by this Agreement.
+
+1. License Rights and Redistribution.
+
+a. Grant of Rights. You are granted a non-exclusive, worldwide, non-transferable and royalty-free limited license under Meta’s intellectual property or other rights owned by Meta embodied in the Llama Materials to use, reproduce, distribute, copy, create derivative works of, and make modifications to the Llama Materials.
+
+b. Redistribution and Use.
+
+i. If you distribute or make the Llama Materials, or any derivative works thereof, available to a third party, you shall provide a copy of this Agreement to such third party.
+
+ii. If you receive Llama Materials, or any derivative works thereof, from a Licensee as part of an integrated end user product, then Section 2 of this Agreement will not apply to you.
+
+iii. You must retain in all copies of the Llama Materials that you distribute the following attribution notice within a “Notice” text file distributed as a part of such copies: “Llama 2 is licensed under the LLAMA 2 Community License, Copyright © Meta Platforms, Inc. All Rights Reserved.”
+
+iv. Your use of the Llama Materials must comply with applicable laws and regulations (including trade compliance laws and regulations) and adhere to the Acceptable Use Policy for the Llama Materials (available at https://ai.meta.com/llama/use-policy), which is hereby incorporated by reference into this Agreement.
+
+v. You will not use the Llama Materials or any output or results of the Llama Materials to improve any other large language model (excluding Llama 2 or derivative works thereof).
+
+2. Additional Commercial Terms. If, on the Llama 2 version release date, the monthly active users of the products or services made available by or for Licensee, or Licensee’s affiliates, is greater than 700 million monthly active users in the preceding calendar month, you must request a license from Meta, which Meta may grant to you in its sole discretion, and you are not authorized to exercise any of the rights under this Agreement unless or until Meta otherwise expressly grants you such rights.
+
+3. Disclaimer of Warranty. UNLESS REQUIRED BY APPLICABLE LAW, THE LLAMA MATERIALS AND ANY OUTPUT AND RESULTS THEREFROM ARE PROVIDED ON AN “AS IS” BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED, INCLUDING, WITHOUT LIMITATION, ANY WARRANTIES OF TITLE, NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE. YOU ARE SOLELY RESPONSIBLE FOR DETERMINING THE APPROPRIATENESS OF USING OR REDISTRIBUTING THE LLAMA MATERIALS AND ASSUME ANY RISKS ASSOCIATED WITH YOUR USE OF THE LLAMA MATERIALS AND ANY OUTPUT AND RESULTS.
+
+4. Limitation of Liability. IN NO EVENT WILL META OR ITS AFFILIATES BE LIABLE UNDER ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, TORT, NEGLIGENCE, PRODUCTS LIABILITY, OR OTHERWISE, ARISING OUT OF THIS AGREEMENT, FOR ANY LOST PROFITS OR ANY INDIRECT, SPECIAL, CONSEQUENTIAL, INCIDENTAL, EXEMPLARY OR PUNITIVE DAMAGES, EVEN IF META OR ITS AFFILIATES HAVE BEEN ADVISED OF THE POSSIBILITY OF ANY OF THE FOREGOING.
+
+5. Intellectual Property.
+
+a. No trademark licenses are granted under this Agreement, and in connection with the Llama Materials, neither Meta nor Licensee may use any name or mark owned by or associated with the other or any of its affiliates, except as required for reasonable and customary use in describing and redistributing the Llama Materials.
+
+b. Subject to Meta’s ownership of Llama Materials and derivatives made by or for Meta, with respect to any derivative works and modifications of the Llama Materials that are made by you, as between you and Meta, you are and will be the owner of such derivative works and modifications.
+
+c. If you institute litigation or other proceedings against Meta or any entity (including a cross-claim or counterclaim in a lawsuit) alleging that the Llama Materials or Llama 2 outputs or results, or any portion of any of the foregoing, constitutes infringement of intellectual property or other rights owned or licensable by you, then any licenses granted to you under this Agreement shall terminate as of the date such litigation or claim is filed or instituted. You will indemnify and hold harmless Meta from and against any claim by any third party arising out of or related to your use or distribution of the Llama Materials.
+
+6. Term and Termination. The term of this Agreement will commence upon your acceptance of this Agreement or access to the Llama Materials and will continue in full force and effect until terminated in accordance with the terms and conditions herein. Meta may terminate this Agreement if you are in breach of any term or condition of this Agreement. Upon termination of this Agreement, you shall delete and cease use of the Llama Materials. Sections 3, 4 and 7 shall survive the termination of this Agreement.
+
+7. Governing Law and Jurisdiction. This Agreement will be governed and construed under the laws of the State of California without regard to choice of law principles, and the UN Convention on Contracts for the International Sale of Goods does not apply to this Agreement. The courts of California shall have exclusive jurisdiction of any dispute arising out of this Agreement.
+
+"""
+
+LICENSE """
+Llama 2 Acceptable Use Policy
+
+Meta is committed to promoting safe and fair use of its tools and features, including Llama 2. If you access or use Llama 2, you agree to this Acceptable Use Policy (“Policy”). The most recent copy of this policy can be found at ai.meta.com/llama/use-policy.
+
+Prohibited Uses
+
+We want everyone to use Llama 2 safely and responsibly. You agree you will not use, or allow others to use, Llama 2 to:
+
+1. Violate the law or others’ rights, including to:
+
+a. Engage in, promote, generate, contribute to, encourage, plan, incite, or further illegal or unlawful activity or content, such as:
+
+i. Violence or terrorism
+
+ii. Exploitation or harm to children, including the solicitation, creation, acquisition, or dissemination of child exploitative content or failure to report Child Sexual Abuse Material
+
+b. Human trafficking, exploitation, and sexual violence
+
+iii. The illegal distribution of information or materials to minors, including obscene materials, or failure to employ legally required age-gating in connection with such information or materials.
+
+iv. Sexual solicitation
+
+vi. Any other criminal activity
+
+c. Engage in, promote, incite, or facilitate the harassment, abuse, threatening, or bullying of individuals or groups of individuals
+
+d. Engage in, promote, incite, or facilitate discrimination or other unlawful or harmful conduct in the provision of employment, employment benefits, credit, housing, other economic benefits, or other essential goods and services
+
+e. Engage in the unauthorized or unlicensed practice of any profession including, but not limited to, financial, legal, medical/health, or related professional practices
+
+f. Collect, process, disclose, generate, or infer health, demographic, or other sensitive personal or private information about individuals without rights and consents required by applicable laws
+
+g. Engage in or facilitate any action or generate any content that infringes, misappropriates, or otherwise violates any third-party rights, including the outputs or results of any products or services using the Llama 2 Materials
+
+h. Create, generate, or facilitate the creation of malicious code, malware, computer viruses or do anything else that could disable, overburden, interfere with or impair the proper working, integrity, operation or appearance of a website or computer system
+
+2. Engage in, promote, incite, facilitate, or assist in the planning or development of activities that present a risk of death or bodily harm to individuals, including use of Llama 2 related to the following:
+
+a. Military, warfare, nuclear industries or applications, espionage, use for materials or activities that are subject to the International Traffic Arms Regulations (ITAR) maintained by the United States Department of State
+
+b. Guns and illegal weapons (including weapon development)
+
+c. Illegal drugs and regulated/controlled substances
+
+d. Operation of critical infrastructure, transportation technologies, or heavy machinery
+
+e. Self-harm or harm to others, including suicide, cutting, and eating disorders
+
+f. Any content intended to incite or promote violence, abuse, or any infliction of bodily harm to an individual
+
+3. Intentionally deceive or mislead others, including use of Llama 2 related to the following:
+
+a. Generating, promoting, or furthering fraud or the creation or promotion of disinformation
+
+b. Generating, promoting, or furthering defamatory content, including the creation of defamatory statements, images, or other content
+
+c. Generating, promoting, or further distributing spam
+
+d. Impersonating another individual without consent, authorization, or legal right
+
+e. Representing that the use of Llama 2 or outputs are human-generated
+
+f. Generating or facilitating false online engagement, including fake reviews and other means of fake online engagement
+
+4. Fail to appropriately disclose to end users any known dangers of your AI system
+
+Please report any violation of this Policy, software “bug,” or other problems that could lead to a violation of this Policy through one of the following means:
+
+Reporting issues with the model: github.com/facebookresearch/llama
+Reporting risky content generated by the model: developers.facebook.com/llama_output_feedback
+Reporting bugs and security concerns: facebook.com/whitehat/info
+Reporting violations of the Acceptable Use Policy or unlicensed uses of Llama: LlamaUseReport@meta.com
+"""

library/modelfiles/llama2_7b

@@ -0,0 +1,147 @@
+FROM ../models/llama-2-7b-chat.ggmlv3.q4_0.bin
+
+TEMPLATE """
+{{- if .First }}
+<<SYS>>
+{{ .System }}
+<</SYS>>
+{{- end }}
+
+[INST] {{ .Prompt }} [/INST]
+"""
+
+SYSTEM """
+You are a helpful, respectful and honest assistant. Always answer as helpfully as possible, while being safe. Your answers should not include any harmful, unethical, racist, sexist, toxic, dangerous, or illegal content. Please ensure that your responses are socially unbiased and positive in nature.
+
+If a question does not make any sense, or is not factually coherent, explain why instead of answering something not correct. If you don't know the answer to a question, please don't share false information.
+"""
+
+LICENSE """
+Llama 2 Community License Agreement
+
+Llama 2 Version Release Date: July 18, 2023
+
+“Agreement” means the terms and conditions for use, reproduction, distribution and modification of the Llama Materials set forth herein.
+
+“Documentation” means the specifications, manuals and documentation accompanying Llama 2 distributed by Meta at ai.meta.com/resources/models-and-libraries/llama-downloads/.
+
+“Licensee” or “you” means you, or your employer or any other person or entity (if you are entering into this Agreement on such person or entity’s behalf), of the age required under applicable laws, rules or regulations to provide legal consent and that has legal authority to bind your employer or such other person or entity if you are entering in this Agreement on their behalf.
+
+“Llama 2” means the foundational large language models and software and algorithms, including machine-learning model code, trained model weights, inference-enabling code, training-enabling code, fine-tuning enabling code and other elements of the foregoing distributed by Meta at ai.meta.com/resources/models-and-libraries/llama-downloads/.
+
+“Llama Materials” means, collectively, Meta’s proprietary Llama 2 and Documentation (and any portion thereof) made available under this Agreement.
+
+“Meta” or “we” means Meta Platforms Ireland Limited (if you are located in or, if you are an entity, your principal place of business is in the EEA or Switzerland) and Meta Platforms, Inc. (if you are located outside of the EEA or Switzerland).
+
+By clicking “I Accept” below or by using or distributing any portion or element of the Llama Materials, you agree to be bound by this Agreement.
+
+1. License Rights and Redistribution.
+
+a. Grant of Rights. You are granted a non-exclusive, worldwide, non-transferable and royalty-free limited license under Meta’s intellectual property or other rights owned by Meta embodied in the Llama Materials to use, reproduce, distribute, copy, create derivative works of, and make modifications to the Llama Materials.
+
+b. Redistribution and Use.
+
+i. If you distribute or make the Llama Materials, or any derivative works thereof, available to a third party, you shall provide a copy of this Agreement to such third party.
+
+ii. If you receive Llama Materials, or any derivative works thereof, from a Licensee as part of an integrated end user product, then Section 2 of this Agreement will not apply to you.
+
+iii. You must retain in all copies of the Llama Materials that you distribute the following attribution notice within a “Notice” text file distributed as a part of such copies: “Llama 2 is licensed under the LLAMA 2 Community License, Copyright © Meta Platforms, Inc. All Rights Reserved.”
+
+iv. Your use of the Llama Materials must comply with applicable laws and regulations (including trade compliance laws and regulations) and adhere to the Acceptable Use Policy for the Llama Materials (available at https://ai.meta.com/llama/use-policy), which is hereby incorporated by reference into this Agreement.
+
+v. You will not use the Llama Materials or any output or results of the Llama Materials to improve any other large language model (excluding Llama 2 or derivative works thereof).
+
+2. Additional Commercial Terms. If, on the Llama 2 version release date, the monthly active users of the products or services made available by or for Licensee, or Licensee’s affiliates, is greater than 700 million monthly active users in the preceding calendar month, you must request a license from Meta, which Meta may grant to you in its sole discretion, and you are not authorized to exercise any of the rights under this Agreement unless or until Meta otherwise expressly grants you such rights.
+
+3. Disclaimer of Warranty. UNLESS REQUIRED BY APPLICABLE LAW, THE LLAMA MATERIALS AND ANY OUTPUT AND RESULTS THEREFROM ARE PROVIDED ON AN “AS IS” BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED, INCLUDING, WITHOUT LIMITATION, ANY WARRANTIES OF TITLE, NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE. YOU ARE SOLELY RESPONSIBLE FOR DETERMINING THE APPROPRIATENESS OF USING OR REDISTRIBUTING THE LLAMA MATERIALS AND ASSUME ANY RISKS ASSOCIATED WITH YOUR USE OF THE LLAMA MATERIALS AND ANY OUTPUT AND RESULTS.
+
+4. Limitation of Liability. IN NO EVENT WILL META OR ITS AFFILIATES BE LIABLE UNDER ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, TORT, NEGLIGENCE, PRODUCTS LIABILITY, OR OTHERWISE, ARISING OUT OF THIS AGREEMENT, FOR ANY LOST PROFITS OR ANY INDIRECT, SPECIAL, CONSEQUENTIAL, INCIDENTAL, EXEMPLARY OR PUNITIVE DAMAGES, EVEN IF META OR ITS AFFILIATES HAVE BEEN ADVISED OF THE POSSIBILITY OF ANY OF THE FOREGOING.
+
+5. Intellectual Property.
+
+a. No trademark licenses are granted under this Agreement, and in connection with the Llama Materials, neither Meta nor Licensee may use any name or mark owned by or associated with the other or any of its affiliates, except as required for reasonable and customary use in describing and redistributing the Llama Materials.
+
+b. Subject to Meta’s ownership of Llama Materials and derivatives made by or for Meta, with respect to any derivative works and modifications of the Llama Materials that are made by you, as between you and Meta, you are and will be the owner of such derivative works and modifications.
+
+c. If you institute litigation or other proceedings against Meta or any entity (including a cross-claim or counterclaim in a lawsuit) alleging that the Llama Materials or Llama 2 outputs or results, or any portion of any of the foregoing, constitutes infringement of intellectual property or other rights owned or licensable by you, then any licenses granted to you under this Agreement shall terminate as of the date such litigation or claim is filed or instituted. You will indemnify and hold harmless Meta from and against any claim by any third party arising out of or related to your use or distribution of the Llama Materials.
+
+6. Term and Termination. The term of this Agreement will commence upon your acceptance of this Agreement or access to the Llama Materials and will continue in full force and effect until terminated in accordance with the terms and conditions herein. Meta may terminate this Agreement if you are in breach of any term or condition of this Agreement. Upon termination of this Agreement, you shall delete and cease use of the Llama Materials. Sections 3, 4 and 7 shall survive the termination of this Agreement.
+
+7. Governing Law and Jurisdiction. This Agreement will be governed and construed under the laws of the State of California without regard to choice of law principles, and the UN Convention on Contracts for the International Sale of Goods does not apply to this Agreement. The courts of California shall have exclusive jurisdiction of any dispute arising out of this Agreement.
+
+"""
+
+LICENSE """
+Llama 2 Acceptable Use Policy
+
+Meta is committed to promoting safe and fair use of its tools and features, including Llama 2. If you access or use Llama 2, you agree to this Acceptable Use Policy (“Policy”). The most recent copy of this policy can be found at ai.meta.com/llama/use-policy.
+
+Prohibited Uses
+
+We want everyone to use Llama 2 safely and responsibly. You agree you will not use, or allow others to use, Llama 2 to:
+
+1. Violate the law or others’ rights, including to:
+
+a. Engage in, promote, generate, contribute to, encourage, plan, incite, or further illegal or unlawful activity or content, such as:
+
+i. Violence or terrorism
+
+ii. Exploitation or harm to children, including the solicitation, creation, acquisition, or dissemination of child exploitative content or failure to report Child Sexual Abuse Material
+
+b. Human trafficking, exploitation, and sexual violence
+
+iii. The illegal distribution of information or materials to minors, including obscene materials, or failure to employ legally required age-gating in connection with such information or materials.
+
+iv. Sexual solicitation
+
+vi. Any other criminal activity
+
+c. Engage in, promote, incite, or facilitate the harassment, abuse, threatening, or bullying of individuals or groups of individuals
+
+d. Engage in, promote, incite, or facilitate discrimination or other unlawful or harmful conduct in the provision of employment, employment benefits, credit, housing, other economic benefits, or other essential goods and services
+
+e. Engage in the unauthorized or unlicensed practice of any profession including, but not limited to, financial, legal, medical/health, or related professional practices
+
+f. Collect, process, disclose, generate, or infer health, demographic, or other sensitive personal or private information about individuals without rights and consents required by applicable laws
+
+g. Engage in or facilitate any action or generate any content that infringes, misappropriates, or otherwise violates any third-party rights, including the outputs or results of any products or services using the Llama 2 Materials
+
+h. Create, generate, or facilitate the creation of malicious code, malware, computer viruses or do anything else that could disable, overburden, interfere with or impair the proper working, integrity, operation or appearance of a website or computer system
+
+2. Engage in, promote, incite, facilitate, or assist in the planning or development of activities that present a risk of death or bodily harm to individuals, including use of Llama 2 related to the following:
+
+a. Military, warfare, nuclear industries or applications, espionage, use for materials or activities that are subject to the International Traffic Arms Regulations (ITAR) maintained by the United States Department of State
+
+b. Guns and illegal weapons (including weapon development)
+
+c. Illegal drugs and regulated/controlled substances
+
+d. Operation of critical infrastructure, transportation technologies, or heavy machinery
+
+e. Self-harm or harm to others, including suicide, cutting, and eating disorders
+
+f. Any content intended to incite or promote violence, abuse, or any infliction of bodily harm to an individual
+
+3. Intentionally deceive or mislead others, including use of Llama 2 related to the following:
+
+a. Generating, promoting, or furthering fraud or the creation or promotion of disinformation
+
+b. Generating, promoting, or furthering defamatory content, including the creation of defamatory statements, images, or other content
+
+c. Generating, promoting, or further distributing spam
+
+d. Impersonating another individual without consent, authorization, or legal right
+
+e. Representing that the use of Llama 2 or outputs are human-generated
+
+f. Generating or facilitating false online engagement, including fake reviews and other means of fake online engagement
+
+4. Fail to appropriately disclose to end users any known dangers of your AI system
+
+Please report any violation of this Policy, software “bug,” or other problems that could lead to a violation of this Policy through one of the following means:
+
+Reporting issues with the model: github.com/facebookresearch/llama
+Reporting risky content generated by the model: developers.facebook.com/llama_output_feedback
+Reporting bugs and security concerns: facebook.com/whitehat/info
+Reporting violations of the Acceptable Use Policy or unlicensed uses of Llama: LlamaUseReport@meta.com
+"""

library/modelfiles/nous-hermes

@@ -0,0 +1,7 @@
+FROM ../models/nous-hermes-13b.ggmlv3.q4_0.bin
+TEMPLATE """
+### Instruction:
+{{ .Prompt }}
+
+### Response:
+"""

library/modelfiles/orca

@@ -0,0 +1,14 @@
+FROM ../models/orca-mini-3b.ggmlv3.q4_0.bin
+TEMPLATE """
+{{- if .First }}
+### System:
+{{ .System }}
+{{- end }}
+
+### User:
+{{ .Prompt }}
+
+### Response:
+"""
+
+SYSTEM """You are an AI assistant that follows instruction extremely well. Help as much as you can."""

library/modelfiles/vicuna

@@ -0,0 +1,11 @@
+FROM ../models/vicuna-7b-v1.3.ggmlv3.q4_0.bin
+TEMPLATE """
+{{ if .First }}
+{{ .System }}
+{{- end }}
+
+USER: {{ .Prompt }}
+ASSISTANT:
+"""
+
+SYSTEM """A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions."""

library/modelfiles/wizard-vicuna

@@ -0,0 +1,5 @@
+FROM ../models/Wizard-Vicuna-13B-Uncensored.ggmlv3.q4_0.bin
+TEMPLATE """
+USER: {{ .Prompt }}
+ASSISTANT:
+"""

library/publish.sh

@@ -0,0 +1,52 @@
+#!/bin/bash
+
+mkdir -p models
+
+# download binaries
+function process_line {
+    local url=$1
+    local checksum=$2
+
+    # Get the filename from the URL
+    local filename=models/$(basename $url)
+
+    echo "verifying $filename..."
+
+    # If the file exists, compute its checksum
+    if [ -f $filename ]; then
+        local existing_checksum=$(shasum -a 256 $filename | cut -d ' ' -f1)
+    fi
+
+    # If the file does not exist, or its checksum does not match, download it
+    if [ ! -f $filename ] || [ $existing_checksum != $checksum ]; then
+        echo "downloading $filename..."
+        
+        # Download the file
+        curl -L $url -o $filename
+
+        # Compute the SHA256 hash of the downloaded file
+        local computed_checksum=$(shasum -a 256 $filename | cut -d ' ' -f1)
+
+        # Verify the checksum
+        if [ $computed_checksum != $checksum ]; then
+            echo "Checksum verification failed for $filename"
+            exit 1
+        fi
+    fi
+}
+
+while IFS=' ' read -r url checksum
+do
+    process_line $url $checksum
+done < "downloads"
+
+# create and publish the models
+for file in modelfiles/*; do
+  if [ -f "$file" ]; then
+    filename=$(basename "$file")
+    echo $filename
+    ollama create "library/${filename}" -f "$file"
+    ollama push "${filename}"
+  fi
+done
+

llama/ggml-cuda.h

@@ -0,0 +1,62 @@
+/**
+ * llama.cpp - git e782c9e735f93ab4767ffc37462c523b73a17ddc
+ *
+ * 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
+
+#define GGML_CUDA_MAX_DEVICES       16
+
+void   ggml_init_cublas(void);
+void   ggml_cuda_set_tensor_split(const float * tensor_split);
+
+void   ggml_cuda_mul(const struct ggml_tensor * src0, const struct ggml_tensor * src1, struct ggml_tensor * dst);
+bool   ggml_cuda_can_mul_mat(const struct ggml_tensor * src0, const struct ggml_tensor * src1, struct ggml_tensor * dst);
+size_t ggml_cuda_mul_mat_get_wsize(const struct ggml_tensor * src0, const struct ggml_tensor * src1, struct ggml_tensor * dst);
+void   ggml_cuda_mul_mat(const struct ggml_tensor * src0, const struct ggml_tensor * src1, struct ggml_tensor * dst, void * wdata, size_t wsize);
+
+// TODO: export these with GGML_API
+void * ggml_cuda_host_malloc(size_t size);
+void   ggml_cuda_host_free(void * ptr);
+
+void   ggml_cuda_transform_tensor(void * data, struct ggml_tensor * tensor);
+
+void   ggml_cuda_free_data(struct ggml_tensor * tensor);
+void   ggml_cuda_assign_buffers(struct ggml_tensor * tensor);
+void   ggml_cuda_assign_buffers_no_scratch(struct ggml_tensor * tensor);
+void   ggml_cuda_assign_buffers_force_inplace(struct ggml_tensor * tensor);
+void   ggml_cuda_set_main_device(int main_device);
+void   ggml_cuda_set_scratch_size(size_t scratch_size);
+void   ggml_cuda_free_scratch(void);
+bool   ggml_cuda_compute_forward(struct ggml_compute_params * params, struct ggml_tensor * tensor);
+
+#ifdef  __cplusplus
+}
+#endif

llama/ggml-metal.h

@@ -0,0 +1,99 @@
+//go:build darwin
+
+/**
+ * llama.cpp - git e782c9e735f93ab4767ffc37462c523b73a17ddc
+ *
+ * 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 <stddef.h>
+#include <stdbool.h>
+
+// max memory buffers that can be mapped to the device
+#define GGML_METAL_MAX_BUFFERS 16
+
+struct ggml_tensor;
+struct ggml_cgraph;
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+struct ggml_metal_context;
+
+// 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);
+
+// 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);
+
+// 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);
+
+#ifdef __cplusplus
+}
+#endif
+

llama/ggml-mpi.c

@@ -0,0 +1,244 @@
+//go:build mpi
+
+/**
+ * llama.cpp - git e782c9e735f93ab4767ffc37462c523b73a17ddc
+ *
+ * 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);
+    }
+}

llama/ggml-mpi.h

@@ -0,0 +1,67 @@
+//go:build mpi
+
+/**
+ * llama.cpp - git e782c9e735f93ab4767ffc37462c523b73a17ddc
+ *
+ * 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

llama/ggml-opencl.h

@@ -0,0 +1,53 @@
+//go:build opencl
+
+/**
+ * llama.cpp - git e782c9e735f93ab4767ffc37462c523b73a17ddc
+ *
+ * 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

llama/k_quants.h

@@ -0,0 +1,191 @@
+/**
+ * llama.cpp - git e782c9e735f93ab4767ffc37462c523b73a17ddc
+ *
+ * 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 elemenets 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 elemenets 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
+// 16 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
+// 16 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 elemenets 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);
+

llama/llama-util.h

@@ -0,0 +1,530 @@
+/**
+ * llama.cpp - git e782c9e735f93ab4767ffc37462c523b73a17ddc
+ *
+ * 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.
+ */
+
+// Internal header to be included only by llama.cpp.
+// Contains wrappers around OS interfaces.
+
+#ifndef LLAMA_UTIL_H
+#define LLAMA_UTIL_H
+
+#include <cstdio>
+#include <cstdint>
+#include <cerrno>
+#include <cstring>
+#include <cstdarg>
+#include <cstdlib>
+#include <climits>
+
+#include <string>
+#include <vector>
+#include <stdexcept>
+
+#ifdef __has_include
+    #if __has_include(<unistd.h>)
+        #include <unistd.h>
+        #if defined(_POSIX_MAPPED_FILES)
+            #include <sys/mman.h>
+        #endif
+        #if defined(_POSIX_MEMLOCK_RANGE)
+            #include <sys/resource.h>
+        #endif
+    #endif
+#endif
+
+#if defined(_WIN32)
+    #define WIN32_LEAN_AND_MEAN
+    #ifndef NOMINMAX
+        #define NOMINMAX
+    #endif
+    #include <windows.h>
+    #include <io.h>
+    #include <stdio.h> // for _fseeki64
+#endif
+
+#define LLAMA_ASSERT(x) \
+    do { \
+        if (!(x)) { \
+            fprintf(stderr, "LLAMA_ASSERT: %s:%d: %s\n", __FILE__, __LINE__, #x); \
+            abort(); \
+        } \
+    } while (0)
+
+#ifdef __GNUC__
+#ifdef __MINGW32__
+__attribute__((format(gnu_printf, 1, 2)))
+#else
+__attribute__((format(printf, 1, 2)))
+#endif
+#endif
+static std::string format(const char * fmt, ...) {
+    va_list ap, ap2;
+    va_start(ap, fmt);
+    va_copy(ap2, ap);
+    int size = vsnprintf(NULL, 0, fmt, ap);
+    LLAMA_ASSERT(size >= 0 && size < INT_MAX);
+    std::vector<char> buf(size + 1);
+    int size2 = vsnprintf(buf.data(), size + 1, fmt, ap2);
+    LLAMA_ASSERT(size2 == size);
+    va_end(ap2);
+    va_end(ap);
+    return std::string(buf.data(), size);
+}
+
+struct llama_file {
+    // use FILE * so we don't have to re-open the file to mmap
+    FILE * fp;
+    size_t size;
+
+    llama_file(const char * fname, const char * mode) {
+        fp = std::fopen(fname, mode);
+        if (fp == NULL) {
+            throw std::runtime_error(format("failed to open %s: %s", fname, strerror(errno)));
+        }
+        seek(0, SEEK_END);
+        size = tell();
+        seek(0, SEEK_SET);
+    }
+
+    size_t tell() const {
+#ifdef _WIN32
+        __int64 ret = _ftelli64(fp);
+#else
+        long ret = std::ftell(fp);
+#endif
+        LLAMA_ASSERT(ret != -1); // this really shouldn't fail
+        return (size_t) ret;
+    }
+
+    void seek(size_t offset, int whence) {
+#ifdef _WIN32
+        int ret = _fseeki64(fp, (__int64) offset, whence);
+#else
+        int ret = std::fseek(fp, (long) offset, whence);
+#endif
+        LLAMA_ASSERT(ret == 0); // same
+    }
+
+    void read_raw(void * ptr, size_t len) const {
+        if (len == 0) {
+            return;
+        }
+        errno = 0;
+        std::size_t ret = std::fread(ptr, len, 1, fp);
+        if (ferror(fp)) {
+            throw std::runtime_error(format("read error: %s", strerror(errno)));
+        }
+        if (ret != 1) {
+            throw std::runtime_error(std::string("unexpectedly reached end of file"));
+        }
+    }
+
+    std::uint32_t read_u32() {
+        std::uint32_t ret;
+        read_raw(&ret, sizeof(ret));
+        return ret;
+    }
+
+    std::string read_string(std::uint32_t len) {
+        std::vector<char> chars(len);
+        read_raw(chars.data(), len);
+        return std::string(chars.data(), len);
+    }
+
+    void write_raw(const void * ptr, size_t len) const {
+        if (len == 0) {
+            return;
+        }
+        errno = 0;
+        size_t ret = std::fwrite(ptr, len, 1, fp);
+        if (ret != 1) {
+            throw std::runtime_error(format("write error: %s", strerror(errno)));
+        }
+    }
+
+    void write_u32(std::uint32_t val) {
+        write_raw(&val, sizeof(val));
+    }
+
+    ~llama_file() {
+        if (fp) {
+            std::fclose(fp);
+        }
+    }
+};
+
+#if defined(_WIN32)
+static std::string llama_format_win_err(DWORD err) {
+    LPSTR buf;
+    size_t size = FormatMessageA(FORMAT_MESSAGE_ALLOCATE_BUFFER | FORMAT_MESSAGE_FROM_SYSTEM | FORMAT_MESSAGE_IGNORE_INSERTS,
+                                 NULL, err, MAKELANGID(LANG_NEUTRAL, SUBLANG_DEFAULT), (LPSTR)&buf, 0, NULL);
+    if (!size) {
+        return "FormatMessageA failed";
+    }
+    std::string ret(buf, size);
+    LocalFree(buf);
+    return ret;
+}
+#endif
+
+struct llama_mmap {
+    void * addr;
+    size_t size;
+
+    llama_mmap(const llama_mmap &) = delete;
+
+#ifdef _POSIX_MAPPED_FILES
+    static constexpr bool SUPPORTED = true;
+
+    llama_mmap(struct llama_file * file, size_t prefetch = (size_t) -1 /* -1 = max value */, bool numa = false) {
+        size = file->size;
+        int fd = fileno(file->fp);
+        int flags = MAP_SHARED;
+        // prefetch/readahead impairs performance on NUMA systems
+        if (numa) { prefetch = 0; }
+#ifdef __linux__
+        if (prefetch) { flags |= MAP_POPULATE; }
+#endif
+        addr = mmap(NULL, file->size, PROT_READ, flags, fd, 0);
+        if (addr == MAP_FAILED) {
+            throw std::runtime_error(format("mmap failed: %s", strerror(errno)));
+        }
+
+        if (prefetch > 0) {
+            // Advise the kernel to preload the mapped memory
+            if (madvise(addr, std::min(file->size, prefetch), MADV_WILLNEED)) {
+                fprintf(stderr, "warning: madvise(.., MADV_WILLNEED) failed: %s\n",
+                        strerror(errno));
+            }
+        }
+        if (numa) {
+            // advise the kernel not to use readahead
+            // (because the next page might not belong on the same node)
+            if (madvise(addr, file->size, MADV_RANDOM)) {
+                fprintf(stderr, "warning: madvise(.., MADV_RANDOM) failed: %s\n",
+                        strerror(errno));
+            }
+        }
+    }
+
+    ~llama_mmap() {
+        munmap(addr, size);
+    }
+#elif defined(_WIN32)
+    static constexpr bool SUPPORTED = true;
+
+    llama_mmap(struct llama_file * file, bool prefetch = true, bool numa = false) {
+        (void) numa;
+
+        size = file->size;
+
+        HANDLE hFile = (HANDLE) _get_osfhandle(_fileno(file->fp));
+
+        HANDLE hMapping = CreateFileMappingA(hFile, NULL, PAGE_READONLY, 0, 0, NULL);
+        DWORD error = GetLastError();
+
+        if (hMapping == NULL) {
+            throw std::runtime_error(format("CreateFileMappingA failed: %s", llama_format_win_err(error).c_str()));
+        }
+
+        addr = MapViewOfFile(hMapping, FILE_MAP_READ, 0, 0, 0);
+        error = GetLastError();
+        CloseHandle(hMapping);
+
+        if (addr == NULL) {
+            throw std::runtime_error(format("MapViewOfFile failed: %s", llama_format_win_err(error).c_str()));
+        }
+
+        #if _WIN32_WINNT >= _WIN32_WINNT_WIN8
+        if (prefetch) {
+            // Advise the kernel to preload the mapped memory
+            WIN32_MEMORY_RANGE_ENTRY range;
+            range.VirtualAddress = addr;
+            range.NumberOfBytes = (SIZE_T)size;
+            if (!PrefetchVirtualMemory(GetCurrentProcess(), 1, &range, 0)) {
+                fprintf(stderr, "warning: PrefetchVirtualMemory failed: %s\n",
+                        llama_format_win_err(GetLastError()).c_str());
+            }
+        }
+        #else
+        #pragma message("warning: You are building for pre-Windows 8; prefetch not supported")
+        #endif // _WIN32_WINNT >= _WIN32_WINNT_WIN8
+    }
+
+    ~llama_mmap() {
+        if (!UnmapViewOfFile(addr)) {
+            fprintf(stderr, "warning: UnmapViewOfFile failed: %s\n",
+                    llama_format_win_err(GetLastError()).c_str());
+        }
+    }
+#else
+    static constexpr bool SUPPORTED = false;
+
+    llama_mmap(struct llama_file *, bool prefetch = true, bool numa = false) {
+        (void) prefetch;
+        (void) numa;
+
+        throw std::runtime_error(std::string("mmap not supported"));
+    }
+#endif
+};
+
+// Represents some region of memory being locked using mlock or VirtualLock;
+// will automatically unlock on destruction.
+struct llama_mlock {
+    void * addr = NULL;
+    size_t size = 0;
+    bool failed_already = false;
+
+    llama_mlock() {}
+    llama_mlock(const llama_mlock &) = delete;
+
+    ~llama_mlock() {
+        if (size) {
+            raw_unlock(addr, size);
+        }
+    }
+
+    void init(void * ptr) {
+        LLAMA_ASSERT(addr == NULL && size == 0);
+        addr = ptr;
+    }
+
+    void grow_to(size_t target_size) {
+        LLAMA_ASSERT(addr);
+        if (failed_already) {
+            return;
+        }
+        size_t granularity = lock_granularity();
+        target_size = (target_size + granularity - 1) & ~(granularity - 1);
+        if (target_size > size) {
+            if (raw_lock((uint8_t *) addr + size, target_size - size)) {
+                size = target_size;
+            } else {
+                failed_already = true;
+            }
+        }
+    }
+
+#ifdef _POSIX_MEMLOCK_RANGE
+    static constexpr bool SUPPORTED = true;
+
+    size_t lock_granularity() {
+        return (size_t) sysconf(_SC_PAGESIZE);
+    }
+
+    #ifdef __APPLE__
+        #define MLOCK_SUGGESTION \
+            "Try increasing the sysctl values 'vm.user_wire_limit' and 'vm.global_user_wire_limit' and/or " \
+            "decreasing 'vm.global_no_user_wire_amount'.  Also try increasing RLIMIT_MLOCK (ulimit -l).\n"
+    #else
+        #define MLOCK_SUGGESTION \
+            "Try increasing RLIMIT_MLOCK ('ulimit -l' as root).\n"
+    #endif
+
+    bool raw_lock(const void * addr, size_t size) {
+        if (!mlock(addr, size)) {
+            return true;
+        } else {
+            char* errmsg = std::strerror(errno);
+            bool suggest = (errno == ENOMEM);
+
+            // Check if the resource limit is fine after all
+            struct rlimit lock_limit;
+            if (suggest && getrlimit(RLIMIT_MEMLOCK, &lock_limit))
+                suggest = false;
+            if (suggest && (lock_limit.rlim_max > lock_limit.rlim_cur + size))
+                suggest = false;
+
+            fprintf(stderr, "warning: failed to mlock %zu-byte buffer (after previously locking %zu bytes): %s\n%s",
+                    size, this->size, errmsg, suggest ? MLOCK_SUGGESTION : "");
+            return false;
+        }
+    }
+
+    #undef MLOCK_SUGGESTION
+
+    void raw_unlock(void * addr, size_t size) {
+        if (munlock(addr, size)) {
+            fprintf(stderr, "warning: failed to munlock buffer: %s\n", std::strerror(errno));
+        }
+    }
+#elif defined(_WIN32)
+    static constexpr bool SUPPORTED = true;
+
+    size_t lock_granularity() {
+        SYSTEM_INFO si;
+        GetSystemInfo(&si);
+        return (size_t) si.dwPageSize;
+    }
+
+    bool raw_lock(void * ptr, size_t len) {
+        for (int tries = 1; ; tries++) {
+            if (VirtualLock(ptr, len)) {
+                return true;
+            }
+            if (tries == 2) {
+                fprintf(stderr, "warning: failed to VirtualLock %zu-byte buffer (after previously locking %zu bytes): %s\n",
+                    len, size, llama_format_win_err(GetLastError()).c_str());
+                return false;
+            }
+
+            // It failed but this was only the first try; increase the working
+            // set size and try again.
+            SIZE_T min_ws_size, max_ws_size;
+            if (!GetProcessWorkingSetSize(GetCurrentProcess(), &min_ws_size, &max_ws_size)) {
+                fprintf(stderr, "warning: GetProcessWorkingSetSize failed: %s\n",
+                        llama_format_win_err(GetLastError()).c_str());
+                return false;
+            }
+            // Per MSDN: "The maximum number of pages that a process can lock
+            // is equal to the number of pages in its minimum working set minus
+            // a small overhead."
+            // Hopefully a megabyte is enough overhead:
+            size_t increment = len + 1048576;
+            // The minimum must be <= the maximum, so we need to increase both:
+            min_ws_size += increment;
+            max_ws_size += increment;
+            if (!SetProcessWorkingSetSize(GetCurrentProcess(), min_ws_size, max_ws_size)) {
+                fprintf(stderr, "warning: SetProcessWorkingSetSize failed: %s\n",
+                        llama_format_win_err(GetLastError()).c_str());
+                return false;
+            }
+        }
+    }
+
+    void raw_unlock(void * ptr, size_t len) {
+        if (!VirtualUnlock(ptr, len)) {
+            fprintf(stderr, "warning: failed to VirtualUnlock buffer: %s\n",
+                    llama_format_win_err(GetLastError()).c_str());
+        }
+    }
+#else
+    static constexpr bool SUPPORTED = false;
+
+    size_t lock_granularity() {
+        return (size_t) 65536;
+    }
+
+    bool raw_lock(const void * addr, size_t len) {
+        fprintf(stderr, "warning: mlock not supported on this system\n");
+        return false;
+    }
+
+    void raw_unlock(const void * addr, size_t len) {}
+#endif
+};
+
+// Replacement for std::vector<uint8_t> that doesn't require zero-initialization.
+struct llama_buffer {
+    uint8_t * addr = NULL;
+    size_t size = 0;
+
+    llama_buffer() = default;
+
+    void resize(size_t len) {
+#ifdef GGML_USE_METAL
+        free(addr);
+        int result = posix_memalign((void **) &addr, getpagesize(), len);
+        if (result == 0) {
+            memset(addr, 0, len);
+        }
+        else {
+            addr = NULL;
+        }
+#else
+        delete[] addr;
+        addr = new uint8_t[len];
+#endif
+        size = len;
+    }
+
+    ~llama_buffer() {
+#ifdef GGML_USE_METAL
+        free(addr);
+#else
+        delete[] addr;
+#endif
+        addr = NULL;
+    }
+
+    // disable copy and move
+    llama_buffer(const llama_buffer&) = delete;
+    llama_buffer(llama_buffer&&) = delete;
+    llama_buffer& operator=(const llama_buffer&) = delete;
+    llama_buffer& operator=(llama_buffer&&) = delete;
+};
+
+#ifdef GGML_USE_CUBLAS
+#include "ggml-cuda.h"
+struct llama_ctx_buffer {
+    uint8_t * addr = NULL;
+    bool is_cuda;
+    size_t size = 0;
+
+    llama_ctx_buffer() = default;
+
+    void resize(size_t size) {
+        free();
+
+        addr = (uint8_t *) ggml_cuda_host_malloc(size);
+        if (addr) {
+            is_cuda = true;
+        }
+        else {
+            // fall back to pageable memory
+            addr = new uint8_t[size];
+            is_cuda = false;
+        }
+        this->size = size;
+    }
+
+    void free() {
+        if (addr) {
+            if (is_cuda) {
+                ggml_cuda_host_free(addr);
+            }
+            else {
+                delete[] addr;
+            }
+        }
+        addr = NULL;
+    }
+
+    ~llama_ctx_buffer() {
+        free();
+    }
+
+    // disable copy and move
+    llama_ctx_buffer(const llama_ctx_buffer&) = delete;
+    llama_ctx_buffer(llama_ctx_buffer&&) = delete;
+    llama_ctx_buffer& operator=(const llama_ctx_buffer&) = delete;
+    llama_ctx_buffer& operator=(llama_ctx_buffer&&) = delete;
+};
+#else
+typedef llama_buffer llama_ctx_buffer;
+#endif
+
+#endif

llama/llama.go

@@ -0,0 +1,282 @@
+package llama
+
+/*
+#cgo CPPFLAGS: -O3 -DNDEBUG=1
+#cgo CXXFLAGS: -std=c++11
+#cgo darwin CPPFLAGS: -DGGML_USE_METAL=1 -DGGML_METAL_NDEBUG=1
+#cgo darwin LDFLAGS: -framework Accelerate -framework Foundation -framework Metal -framework MetalKit -framework MetalPerformanceShaders
+#include <stdlib.h>
+#include "llama.h"
+
+struct llama_sample_options
+{
+	float repeat_penalty;
+	float frequency_penalty;
+	float presence_penalty;
+	float temperature;
+	int32_t top_k;
+	float top_p;
+	float tfs_z;
+	float typical_p;
+	int mirostat;
+	float mirostat_tau;
+	float mirostat_eta;
+};
+
+llama_token llama_sample(
+		struct llama_context *ctx,
+		struct llama_token_data *candidates,
+		size_t n_candidates,
+		const llama_token *last_tokens,
+		size_t n_last_tokens,
+		struct llama_sample_options *opts)
+{
+	llama_token_data_array candidates_p = {
+		candidates,
+		n_candidates,
+		false,
+	};
+
+	llama_sample_repetition_penalty(
+		ctx, &candidates_p,
+		last_tokens, n_last_tokens,
+		opts->repeat_penalty);
+
+	llama_sample_frequency_and_presence_penalties(
+		ctx, &candidates_p,
+		last_tokens, n_last_tokens,
+		opts->frequency_penalty, opts->presence_penalty);
+
+	if (opts->temperature <= 0) {
+		return llama_sample_token_greedy(ctx, &candidates_p);
+	}
+
+	if (opts->mirostat == 1) {
+		int mirostat_m = 100;
+		float mirostat_mu = 2.0f * opts->mirostat_tau;
+		llama_sample_temperature(ctx, &candidates_p, opts->temperature);
+		return llama_sample_token_mirostat(
+			ctx, &candidates_p,
+			opts->mirostat_tau, opts->mirostat_eta,
+			mirostat_m, &mirostat_mu);
+	} else if (opts->mirostat == 2) {
+		float mirostat_mu = 2.0f * opts->mirostat_tau;
+		llama_sample_temperature(ctx, &candidates_p, opts->temperature);
+		return llama_sample_token_mirostat_v2(
+			ctx, &candidates_p,
+			opts->mirostat_tau, opts->mirostat_eta,
+			&mirostat_mu);
+	} else {
+		llama_sample_top_k(ctx, &candidates_p, opts->top_k, 1);
+		llama_sample_tail_free(ctx, &candidates_p, opts->tfs_z, 1);
+		llama_sample_typical(ctx, &candidates_p, opts->typical_p, 1);
+		llama_sample_top_p(ctx, &candidates_p, opts->top_p, 1);
+		llama_sample_temperature(ctx, &candidates_p, opts->temperature);
+		return llama_sample_token(ctx, &candidates_p);
+	}
+}
+*/
+import "C"
+import (
+	"bytes"
+	"errors"
+	"fmt"
+	"io"
+	"os"
+	"strings"
+	"time"
+	"unicode/utf8"
+	"unsafe"
+
+	"github.com/jmorganca/ollama/api"
+)
+
+type llama struct {
+	params *C.struct_llama_context_params
+	model  *C.struct_llama_model
+	ctx    *C.struct_llama_context
+
+	api.Options
+}
+
+func New(model string, opts api.Options) (*llama, error) {
+	if _, err := os.Stat(model); err != nil {
+		return nil, err
+	}
+
+	llm := llama{Options: opts}
+
+	C.llama_backend_init(C.bool(llm.UseNUMA))
+
+	params := C.llama_context_default_params()
+	params.seed = C.uint(llm.Seed)
+	params.n_ctx = C.int(llm.NumCtx)
+	params.n_batch = C.int(llm.NumBatch)
+	params.n_gpu_layers = C.int(llm.NumGPU)
+	params.main_gpu = C.int(llm.MainGPU)
+	params.low_vram = C.bool(llm.LowVRAM)
+	params.f16_kv = C.bool(llm.F16KV)
+	params.logits_all = C.bool(llm.LogitsAll)
+	params.vocab_only = C.bool(llm.VocabOnly)
+	params.use_mmap = C.bool(llm.UseMMap)
+	params.use_mlock = C.bool(llm.UseMLock)
+	params.embedding = C.bool(llm.EmbeddingOnly)
+	llm.params = &params
+
+	cModel := C.CString(model)
+	defer C.free(unsafe.Pointer(cModel))
+
+	llm.model = C.llama_load_model_from_file(cModel, params)
+	if llm.model == nil {
+		return nil, errors.New("failed to load model")
+	}
+
+	llm.ctx = C.llama_new_context_with_model(llm.model, params)
+	if llm.ctx == nil {
+		return nil, errors.New("failed to create context")
+	}
+
+	// warm up the model
+	bos := []C.llama_token{C.llama_token_bos()}
+	C.llama_eval(llm.ctx, unsafe.SliceData(bos), C.int(len(bos)), 0, C.int(opts.NumThread))
+	C.llama_reset_timings(llm.ctx)
+
+	return &llm, nil
+}
+
+func (llm *llama) Close() {
+	defer C.llama_free_model(llm.model)
+	defer C.llama_free(llm.ctx)
+
+	C.llama_print_timings(llm.ctx)
+}
+
+func (llm *llama) Predict(ctx []int, prompt string, fn func(api.GenerateResponse)) error {
+	if input := llm.tokenize(prompt); input != nil {
+		embd := make([]C.llama_token, len(ctx))
+		for i := range ctx {
+			embd[i] = C.llama_token(ctx[i])
+		}
+
+		return llm.generate(append(embd, input...), fn)
+	}
+
+	return errors.New("llama: tokenize")
+}
+
+func (llm *llama) tokenize(prompt string) []C.llama_token {
+	cPrompt := C.CString(prompt)
+	defer C.free(unsafe.Pointer(cPrompt))
+
+	tokens := make([]C.llama_token, llm.NumCtx)
+	if n := C.llama_tokenize(llm.ctx, cPrompt, unsafe.SliceData(tokens), C.int(len(tokens)), true); n > 0 {
+		return tokens[:n]
+	}
+
+	return nil
+}
+
+func (llm *llama) detokenize(tokens ...C.llama_token) string {
+	var sb strings.Builder
+	for _, token := range tokens {
+		sb.WriteString(C.GoString(C.llama_token_to_str(llm.ctx, token)))
+	}
+
+	return sb.String()
+}
+
+func (llm *llama) generate(input []C.llama_token, fn func(api.GenerateResponse)) error {
+	var opts C.struct_llama_sample_options
+	opts.repeat_penalty = C.float(llm.RepeatPenalty)
+	opts.frequency_penalty = C.float(llm.FrequencyPenalty)
+	opts.presence_penalty = C.float(llm.PresencePenalty)
+	opts.temperature = C.float(llm.Temperature)
+	opts.top_k = C.int(llm.TopK)
+	opts.top_p = C.float(llm.TopP)
+	opts.tfs_z = C.float(llm.TFSZ)
+	opts.typical_p = C.float(llm.TypicalP)
+	opts.mirostat = C.int(llm.Mirostat)
+	opts.mirostat_tau = C.float(llm.MirostatTau)
+	opts.mirostat_eta = C.float(llm.MirostatEta)
+
+	output := deque[C.llama_token]{capacity: llm.NumCtx}
+
+	context := deque[int]{capacity: llm.NumCtx / 2}
+	for _, in := range input {
+		context.PushLeft(int(in))
+	}
+
+	var b bytes.Buffer
+	for C.llama_get_kv_cache_token_count(llm.ctx) < C.int(llm.NumCtx) {
+		if retval := C.llama_eval(llm.ctx, unsafe.SliceData(input), C.int(len(input)), C.llama_get_kv_cache_token_count(llm.ctx), C.int(llm.NumThread)); retval != 0 {
+			return errors.New("llama: eval")
+		}
+
+		token, err := llm.sample(output, &opts)
+		if errors.Is(err, io.EOF) {
+			break
+		} else if err != nil {
+			return err
+		}
+
+		b.WriteString(llm.detokenize(token))
+		if utf8.Valid(b.Bytes()) || b.Len() >= utf8.UTFMax {
+			// call the callback
+			fn(api.GenerateResponse{
+				Response: b.String(),
+			})
+
+			output.PushLeft(token)
+			context.PushLeft(int(token))
+			b.Reset()
+		}
+
+		input = []C.llama_token{token}
+	}
+
+	dur := func(ms float64) time.Duration {
+		d, err := time.ParseDuration(fmt.Sprintf("%fms", ms))
+		if err != nil {
+			panic(err)
+		}
+
+		return d
+	}
+
+	timings := C.llama_get_timings(llm.ctx)
+	fn(api.GenerateResponse{
+		Done:               true,
+		Context:            context.Data(),
+		PromptEvalCount:    int(timings.n_p_eval),
+		PromptEvalDuration: dur(float64(timings.t_p_eval_ms)),
+		EvalCount:          int(timings.n_eval),
+		EvalDuration:       dur(float64(timings.t_eval_ms)),
+	})
+
+	return nil
+}
+
+func (llm *llama) sample(output deque[C.llama_token], opts *C.struct_llama_sample_options) (C.llama_token, error) {
+	numVocab := int(C.llama_n_vocab(llm.ctx))
+	logits := unsafe.Slice(C.llama_get_logits(llm.ctx), numVocab)
+
+	candidates := deque[C.struct_llama_token_data]{capacity: numVocab}
+	for i := 0; i < candidates.Cap(); i++ {
+		candidates.PushLeft(C.struct_llama_token_data{
+			id:    C.int(i),
+			logit: logits[i],
+			p:     0,
+		})
+	}
+
+	token := C.llama_sample(
+		llm.ctx,
+		unsafe.SliceData(candidates.Data()), C.size_t(candidates.Len()),
+		unsafe.SliceData(output.Data()), C.size_t(output.Len()),
+		opts)
+	if token != C.llama_token_eos() {
+		return token, nil
+	}
+
+	return 0, io.EOF
+}

llama/llama.h

@@ -0,0 +1,440 @@
+/**
+ * llama.cpp - git e782c9e735f93ab4767ffc37462c523b73a17ddc
+ *
+ * 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 <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_FILE_MAGIC_GGJT        0x67676a74u // 'ggjt'
+#define LLAMA_FILE_MAGIC_GGLA        0x67676c61u // 'ggla'
+#define LLAMA_FILE_MAGIC_GGMF        0x67676d66u // 'ggmf'
+#define LLAMA_FILE_MAGIC_GGML        0x67676d6cu // 'ggml'
+#define LLAMA_FILE_MAGIC_GGSN        0x6767736eu // 'ggsn'
+
+#define LLAMA_FILE_VERSION           3
+#define LLAMA_FILE_MAGIC             LLAMA_FILE_MAGIC_GGJT
+#define LLAMA_FILE_MAGIC_UNVERSIONED LLAMA_FILE_MAGIC_GGML
+#define LLAMA_SESSION_MAGIC          LLAMA_FILE_MAGIC_GGSN
+#define LLAMA_SESSION_VERSION        1
+
+#define LLAMA_DEFAULT_SEED           0xFFFFFFFF
+
+#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 int llama_token;
+
+    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);
+
+   struct llama_context_params {
+        uint32_t seed;                         // RNG seed, -1 for random
+        int32_t  n_ctx;                        // text context
+        int32_t  n_batch;                      // prompt processing batch size
+        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
+        float tensor_split[LLAMA_MAX_DEVICES]; // how to split layers across multiple GPUs
+
+        // ref: https://github.com/ggerganov/llama.cpp/pull/2054
+        float    rope_freq_base;  // RoPE base frequency
+        float    rope_freq_scale; // RoPE frequency scaling factor
+
+        // 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 low_vram;   // if true, reduce VRAM usage at the cost of performance
+        bool f16_kv;     // use fp16 for KV cache
+        bool logits_all; // the llama_eval() call computes all logits, not just the last one
+        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
+        bool embedding;  // embedding mode only
+    };
+    // 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
+    };
+
+    // 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
+    } llama_model_quantize_params;
+
+    // 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;
+    };
+
+    LLAMA_API int llama_max_devices();
+
+    LLAMA_API struct llama_context_params llama_context_default_params();
+    LLAMA_API struct llama_model_quantize_params llama_model_quantize_default_params();
+
+    LLAMA_API bool llama_mmap_supported();
+    LLAMA_API bool llama_mlock_supported();
+
+    // TODO: not great API - very likely to change
+    // 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();
+
+    LLAMA_API int64_t llama_time_us();
+
+    LLAMA_API struct llama_model * llama_load_model_from_file(
+                             const char * path_model,
+            struct llama_context_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);
+
+    // Various functions for loading a ggml llama model.
+    // Allocate (almost) all memory needed for the model.
+    // Return NULL on failure
+    LLAMA_API DEPRECATED(struct llama_context * llama_init_from_file(
+                             const char * path_model,
+            struct llama_context_params   params),
+            "please use llama_load_model_from_file combined with llama_new_context_with_model instead");
+
+    // Frees all allocated memory
+    LLAMA_API void llama_free(struct llama_context * ctx);
+
+    // 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,
+                      const char * path_base_model,
+                             int   n_threads),
+            "please 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,
+                      const char * path_base_model,
+                             int   n_threads);
+
+    // Returns the number of tokens in the KV cache
+    LLAMA_API int llama_get_kv_cache_token_count(const struct llama_context * ctx);
+
+    // Sets the current rng seed.
+    LLAMA_API void llama_set_rng_seed(struct llama_context * ctx, uint32_t seed);
+
+    // 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);
+
+    // Run the llama inference to obtain the logits and probabilities for the next token.
+    // 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
+    LLAMA_API int llama_eval(
+            struct llama_context * ctx,
+               const llama_token * tokens,
+                             int   n_tokens,
+                             int   n_past,
+                             int   n_threads);
+
+    // Same as llama_eval, but use float matrix input directly.
+    LLAMA_API int llama_eval_embd(
+            struct llama_context * ctx,
+                     const float * embd,
+                             int   n_tokens,
+                             int   n_past,
+                             int   n_threads);
+
+    // Export a static computation graph for context of 511 and batch size of 1
+    // NOTE: since this functionality is mostly for debugging and demonstration purposes, we hardcode these
+    //       parameters here to keep things simple
+    // IMPORTANT: do not use for anything else other than debugging and testing!
+    LLAMA_API int llama_eval_export(struct llama_context * ctx, const char * fname);
+
+    // Convert the provided text into tokens.
+    // The tokens pointer must be large enough to hold the resulting tokens.
+    // Returns the number of tokens on success, no more than n_max_tokens
+    // Returns a negative number on failure - the number of tokens that would have been returned
+    // TODO: not sure if correct
+    LLAMA_API int llama_tokenize(
+            struct llama_context * ctx,
+                      const char * text,
+                     llama_token * tokens,
+                             int   n_max_tokens,
+                            bool   add_bos);
+
+    LLAMA_API int llama_tokenize_with_model(
+        const struct llama_model * model,
+                      const char * text,
+                     llama_token * tokens,
+                             int   n_max_tokens,
+                            bool   add_bos);
+
+    LLAMA_API int llama_n_vocab(const struct llama_context * ctx);
+    LLAMA_API int llama_n_ctx  (const struct llama_context * ctx);
+    LLAMA_API int llama_n_embd (const struct llama_context * ctx);
+
+    LLAMA_API int llama_n_vocab_from_model(const struct llama_model * model);
+    LLAMA_API int llama_n_ctx_from_model  (const struct llama_model * model);
+    LLAMA_API int llama_n_embd_from_model (const struct llama_model * model);
+
+    // Get the vocabulary as output parameters.
+    // Returns number of results.
+    LLAMA_API int llama_get_vocab(
+            const struct llama_context * ctx,
+                          const char * * strings,
+                                 float * scores,
+                                   int   capacity);
+
+    LLAMA_API int llama_get_vocab_from_model(
+              const struct llama_model * model,
+                          const char * * strings,
+                                 float * scores,
+                                   int   capacity);
+
+    // Token logits obtained from the last call to llama_eval()
+    // The logits for the last token are stored in the last row
+    // Can be mutated in order to change the probabilities of the next token
+    // Rows: n_tokens
+    // Cols: n_vocab
+    LLAMA_API float * llama_get_logits(struct llama_context * ctx);
+
+    // Get the embeddings for the input
+    // shape: [n_embd] (1-dimensional)
+    LLAMA_API float * llama_get_embeddings(struct llama_context * ctx);
+
+    // Token Id -> String. Uses the vocabulary in the provided context
+    LLAMA_API const char * llama_token_to_str(
+            const struct llama_context * ctx,
+                           llama_token   token);
+
+    LLAMA_API const char * llama_token_to_str_with_model(
+              const struct llama_model * model,
+                           llama_token   token);
+
+    // Special tokens
+    LLAMA_API llama_token llama_token_bos();  // beginning-of-sentence
+    LLAMA_API llama_token llama_token_eos();  // end-of-sentence
+    LLAMA_API llama_token llama_token_nl();   // next-line
+
+    // Sampling functions
+
+    /// @details Repetition penalty described in CTRL academic paper https://arxiv.org/abs/1909.05858, with negative logit fix.
+    LLAMA_API void llama_sample_repetition_penalty(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens, size_t last_tokens_size, float penalty);
+
+    /// @details Frequency and presence penalties described in OpenAI API https://platform.openai.com/docs/api-reference/parameter-details.
+    LLAMA_API void llama_sample_frequency_and_presence_penalties(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens, size_t last_tokens_size, float alpha_frequency, float alpha_presence);
+
+    /// @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.
+    /// @params smooth_factor Smooth factor between guidance logits and original logits. 1.0f means only use guidance logits. 0.0f means only original logits.
+    LLAMA_API void llama_sample_classifier_free_guidance(
+              struct llama_context * ctx,
+            llama_token_data_array * candidates,
+              struct llama_context * guidance_ctx,
+                             float   scale,
+                             float   smooth_factor);
+
+    /// @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_temperature(struct llama_context * ctx, llama_token_data_array * candidates, float temp);
+
+    /// @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);
+
+    // 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);
+
+#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
+
+#endif // LLAMA_H

llama/update-llama-cpp.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.*

llama/utils.go

@@ -0,0 +1,104 @@
+package llama
+
+type node[T any] struct {
+	t    T
+	next *node[T]
+	prev *node[T]
+}
+
+type deque[T any] struct {
+	head     *node[T]
+	tail     *node[T]
+	size     int
+	capacity int
+}
+
+func (d *deque[T]) Empty() bool {
+	return d.size == 0
+}
+
+func (d *deque[T]) Len() int {
+	return d.size
+}
+
+func (d *deque[T]) Cap() int {
+	return d.capacity
+}
+
+func (d *deque[T]) Push(t T) {
+	if d.capacity > 0 && d.size >= d.capacity {
+		d.PopLeft()
+	}
+
+	n := node[T]{t: t}
+	if d.head != nil {
+		n.next = d.head
+		d.head.prev = &n
+		d.head = &n
+	} else {
+		d.head = &n
+		d.tail = &n
+	}
+
+	d.size++
+}
+
+func (d *deque[T]) PushLeft(t T) {
+	if d.capacity > 0 && d.size >= d.capacity {
+		d.Pop()
+	}
+
+	n := node[T]{t: t}
+	if d.tail != nil {
+		n.prev = d.tail
+		d.tail.next = &n
+		d.tail = &n
+	} else {
+		d.head = &n
+		d.tail = &n
+	}
+
+	d.size++
+}
+
+func (d *deque[T]) Pop() *T {
+	if d.Empty() {
+		return nil
+	}
+
+	head := d.head
+	d.head = head.next
+	if d.head != nil {
+		d.head.prev = nil
+	} else {
+		d.tail = nil
+	}
+
+	d.size--
+	return &head.t
+}
+
+func (d *deque[T]) PopLeft() *T {
+	if d.Empty() {
+		return nil
+	}
+
+	tail := d.tail
+	d.tail = tail.prev
+	if d.tail != nil {
+		d.tail.next = nil
+	} else {
+		d.head = nil
+	}
+
+	d.size--
+	return &tail.t
+}
+
+func (d *deque[T]) Data() (data []T) {
+	for n := d.head; n != nil; n = n.next {
+		data = append(data, n.t)
+	}
+
+	return data
+}