add recommended models

custom_components/llama_conversation/config_flow.py

@@ -112,6 +112,7 @@ from .const import (
     DOMAIN,
     DEFAULT_OPTIONS,
     OPTIONS_OVERRIDES,
+    RECOMMENDED_CHAT_MODELS,
 )
 
 _LOGGER = logging.getLogger(__name__)
@@ -150,7 +151,12 @@ def STEP_LOCAL_SETUP_EXISTING_DATA_SCHEMA(model_file=None):
 def STEP_LOCAL_SETUP_DOWNLOAD_DATA_SCHEMA(*, chat_model=None, downloaded_model_quantization=None):
     return vol.Schema(
         {
-            vol.Required(CONF_CHAT_MODEL, default=chat_model if chat_model else DEFAULT_CHAT_MODEL): str,
+            vol.Required(CONF_CHAT_MODEL, default=chat_model if chat_model else DEFAULT_CHAT_MODEL): SelectSelector(SelectSelectorConfig(
+                options=RECOMMENDED_CHAT_MODELS,
+                custom_value=True,
+                multiple=False,
+                mode=SelectSelectorMode.DROPDOWN,
+            )),
             vol.Required(CONF_DOWNLOADED_MODEL_QUANTIZATION, default=downloaded_model_quantization if downloaded_model_quantization else DEFAULT_DOWNLOADED_MODEL_QUANTIZATION): vol.In(CONF_DOWNLOADED_MODEL_QUANTIZATION_OPTIONS),
         }
     )

custom_components/llama_conversation/const.py

@@ -25,6 +25,7 @@ Devices:
 {{ devices }}"""
 CONF_CHAT_MODEL = "huggingface_model"
 DEFAULT_CHAT_MODEL = "acon96/Home-3B-v3-GGUF"
+RECOMMENDED_CHAT_MODELS = [ "acon96/Home-3B-v3-GGUF", "acon96/Home-1B-v2-GGUF", "TheBloke/Mistral-7B-Instruct-v0.2-GGUF" ]
 CONF_MAX_TOKENS = "max_new_tokens"
 DEFAULT_MAX_TOKENS = 128
 CONF_TOP_K = "top_k"

custom_components/llama_conversation/utils.py

@@ -68,6 +68,9 @@ def install_llama_cpp_python(config_dir: str):
 
     if is_installed("llama-cpp-python"):
         _LOGGER.info("llama-cpp-python is already installed")
+
+        # not sure why this is still needed
+        time.sleep(0.1)
         return True
     
     platform_suffix = platform.machine()