github/home-llm
1
1
Fork 0
mirror of https://github.com/acon96/home-llm.git synced 2026-01-08 05:14:02 -05:00
Code Issues Packages Projects Releases Wiki Activity

add defaults for functiongemma

Browse Source
This commit is contained in:
Alex O'Connell
2025-12-20 22:41:26 -05:00
parent 0e4031ef43
commit 0b776c0a23
2 changed files with 20 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
TODO.md
View File

@@ -1,7 +1,8 @@
# TODO
- [ ] new model based on qwen3 0.6b, 1.7b and 4b
- [ ] new model based on gemma3 270m
- [ ] support AI task API
- [ ] new model based on qwen3 0.6b, 1.7b and 4b
- [ ] add examples of 'fixing' a failed tool call to the dataset
- [x] new model based on gemma3 270m
- [x] support AI task API
- [ ] vision support for remote backends
- [ ] vision support for local backend (llama.cpp + llava)
- [ ] move llamacpp to a separate process because of all the crashing

30
custom_components/llama_conversation/const.py
View File

@@ -63,7 +63,6 @@ USER_INSTRUCTION = {
"pl": "Instrukcja użytkownika"
}
DEFAULT_PROMPT_BASE = """<persona>
<current_date>
<devices>:
{% for device in devices | selectattr('area_id', 'none'): %}
{{ device.entity_id }} '{{ device.name }}' = {{ device.state }}{{ ([""] + device.attributes) | join(";") }}
@@ -73,23 +72,19 @@ DEFAULT_PROMPT_BASE = """<persona>
{% for device in area.list %}
{{ device.entity_id }} '{{ device.name }}' = {{ device.state }};{{ device.attributes | join(";") }}
{% endfor %}
{% endfor %}"""
{% endfor %}
<current_date>"""
DEFAULT_PROMPT_BASE_LEGACY = """<persona>
<current_date>
<devices>:
{{ formatted_devices }}"""
{{ formatted_devices }}
<current_date>"""
ICL_EXTRAS = """
{% for item in response_examples %}
{{ item.request }}
{{ item.response }}
{{ tool_call_prefix }}{{ item.tool | to_json }}{{ tool_call_suffix }}
{% endfor %}"""
ICL_NO_SYSTEM_PROMPT_EXTRAS = """
{% for item in response_examples %}
{{ item.request }}
{{ item.response }}
{{ tool_call_prefix }}{{ item.tool | to_json }}{{ tool_call_suffix }}
{% endfor %}
NO_SYSTEM_PROMPT_EXTRAS = """
<user_instruction>:"""
DEFAULT_PROMPT = DEFAULT_PROMPT_BASE + ICL_EXTRAS
CONF_CHAT_MODEL = "huggingface_model"
@@ -239,6 +234,13 @@ DEFAULT_OPTIONS = types.MappingProxyType(
def option_overrides(backend_type: str) -> dict[str, Any]:
return {
"home-functiongemma": {
CONF_PROMPT: DEFAULT_PROMPT_BASE + NO_SYSTEM_PROMPT_EXTRAS,
CONF_USE_IN_CONTEXT_LEARNING_EXAMPLES: False,
CONF_TOOL_CALL_PREFIX: "<start_function_call>",
CONF_TOOL_CALL_SUFFIX: "<end_function_call>",
CONF_TOOL_RESPONSE_AS_STRING: False, # gemma function calling requires tool responses as a dictionary
},
"home-llama-3.2": {
CONF_PROMPT: DEFAULT_PROMPT_BASE_LEGACY,
CONF_USE_IN_CONTEXT_LEARNING_EXAMPLES: False,
@@ -247,7 +249,7 @@ def option_overrides(backend_type: str) -> dict[str, Any]:
CONF_CONTEXT_LENGTH: 131072,
CONF_MAX_TOOL_CALL_ITERATIONS: 0,
# llama cpp server doesn't support custom tool calling formats. so just use legacy tool calling
CONF_ENABLE_LEGACY_TOOL_CALLING: backend_type == BACKEND_TYPE_LLAMA_CPP_SERVER
CONF_ENABLE_LEGACY_TOOL_CALLING: True
},
"home-3b-v3": {
CONF_PROMPT: DEFAULT_PROMPT_BASE_LEGACY,
@@ -256,7 +258,7 @@ def option_overrides(backend_type: str) -> dict[str, Any]:
CONF_TOOL_CALL_SUFFIX: "```",
CONF_MAX_TOOL_CALL_ITERATIONS: 0,
# llama cpp server doesn't support custom tool calling formats. so just use legacy tool calling
CONF_ENABLE_LEGACY_TOOL_CALLING: backend_type == BACKEND_TYPE_LLAMA_CPP_SERVER
CONF_ENABLE_LEGACY_TOOL_CALLING: True
},
"home-3b-v2": {
CONF_PROMPT: DEFAULT_PROMPT_BASE_LEGACY,
@@ -310,14 +312,14 @@ def option_overrides(backend_type: str) -> dict[str, Any]:
CONF_TOP_P: 0.95
},
"mistral": {
CONF_PROMPT: DEFAULT_PROMPT_BASE + ICL_NO_SYSTEM_PROMPT_EXTRAS,
CONF_PROMPT: DEFAULT_PROMPT_BASE + ICL_EXTRAS + NO_SYSTEM_PROMPT_EXTRAS,
CONF_MIN_P: 0.1,
CONF_TYPICAL_P: 0.9,
# no prompt formats with tool calling support, so just use legacy tool calling
CONF_ENABLE_LEGACY_TOOL_CALLING: True,
},
"mixtral": {
CONF_PROMPT: DEFAULT_PROMPT_BASE + ICL_NO_SYSTEM_PROMPT_EXTRAS,
CONF_PROMPT: DEFAULT_PROMPT_BASE + ICL_EXTRAS + NO_SYSTEM_PROMPT_EXTRAS,
CONF_MIN_P: 0.1,
CONF_TYPICAL_P: 0.9,
# no prompt formats with tool calling support, so just use legacy tool calling