Add configurable keep_alive to Ollama integration, change default to 5m (#119341)

Co-authored-by: Franck Nijhof <git@frenck.dev>

homeassistant/components/ollama/__init__.py

@@ -14,7 +14,14 @@ from homeassistant.core import HomeAssistant
 from homeassistant.exceptions import ConfigEntryNotReady
 from homeassistant.helpers import config_validation as cv
 
-from .const import CONF_MAX_HISTORY, CONF_MODEL, CONF_PROMPT, DEFAULT_TIMEOUT, DOMAIN
+from .const import (
+    CONF_KEEP_ALIVE,
+    CONF_MAX_HISTORY,
+    CONF_MODEL,
+    CONF_PROMPT,
+    DEFAULT_TIMEOUT,
+    DOMAIN,
+)
 
 _LOGGER = logging.getLogger(__name__)
 
@@ -23,6 +30,7 @@ __all__ = [
     "CONF_PROMPT",
     "CONF_MODEL",
     "CONF_MAX_HISTORY",
+    "CONF_KEEP_ALIVE",
     "DOMAIN",
 ]
 

homeassistant/components/ollama/config_flow.py

@@ -33,9 +33,11 @@ from homeassistant.helpers.selector import (
 )
 
 from .const import (
+    CONF_KEEP_ALIVE,
     CONF_MAX_HISTORY,
     CONF_MODEL,
     CONF_PROMPT,
+    DEFAULT_KEEP_ALIVE,
     DEFAULT_MAX_HISTORY,
     DEFAULT_MODEL,
     DEFAULT_PROMPT,
@@ -235,6 +237,16 @@ def ollama_config_option_schema(options: MappingProxyType[str, Any]) -> dict:
                 min=0, max=sys.maxsize, step=1, mode=NumberSelectorMode.BOX
             )
         ),
+        vol.Optional(
+            CONF_KEEP_ALIVE,
+            description={
+                "suggested_value": options.get(CONF_KEEP_ALIVE, DEFAULT_KEEP_ALIVE)
+            },
+        ): NumberSelector(
+            NumberSelectorConfig(
+                min=-1, max=sys.maxsize, step=1, mode=NumberSelectorMode.BOX
+            )
+        ),
     }
 
 

homeassistant/components/ollama/const.py

@@ -72,6 +72,9 @@ An overview of the areas and the devices in this smart home:
 Answer the user's questions using the information about this smart home.
 Keep your answers brief and do not apologize."""
 
+CONF_KEEP_ALIVE = "keep_alive"
+DEFAULT_KEEP_ALIVE = -1  # seconds. -1 = indefinite, 0 = never
+
 KEEP_ALIVE_FOREVER = -1
 DEFAULT_TIMEOUT = 5.0  # seconds
 

homeassistant/components/ollama/conversation.py

@@ -26,13 +26,14 @@ from homeassistant.helpers.entity_platform import AddEntitiesCallback
 from homeassistant.util import ulid
 
 from .const import (
+    CONF_KEEP_ALIVE,
     CONF_MAX_HISTORY,
     CONF_MODEL,
     CONF_PROMPT,
+    DEFAULT_KEEP_ALIVE,
     DEFAULT_MAX_HISTORY,
     DEFAULT_PROMPT,
     DOMAIN,
-    KEEP_ALIVE_FOREVER,
     MAX_HISTORY_SECONDS,
 )
 from .models import ExposedEntity, MessageHistory, MessageRole
@@ -151,7 +152,8 @@ class OllamaConversationEntity(
                 # Make a copy of the messages because we mutate the list later
                 messages=list(message_history.messages),
                 stream=False,
-                keep_alive=KEEP_ALIVE_FOREVER,
+                # keep_alive requires specifying unit. In this case, seconds
+                keep_alive=f"{settings.get(CONF_KEEP_ALIVE, DEFAULT_KEEP_ALIVE)}s",
             )
         except (ollama.RequestError, ollama.ResponseError) as err:
             _LOGGER.error("Unexpected error talking to Ollama server: %s", err)

homeassistant/components/ollama/strings.json

@@ -25,7 +25,11 @@
       "init": {
         "data": {
           "prompt": "Prompt template",
-          "max_history": "Max history messages"
+          "max_history": "Max history messages",
+          "keep_alive": "Keep alive"
+        },
+        "data_description": {
+          "keep_alive": "Duration in seconds for Ollama to keep model in memory. -1 = indefinite, 0 = never."
         }
       }
     }