65 lines (52 loc) · 6.19 KB

Settings Reference

There are user-side explanations about the settings, and more detailed ones for developers.

LLM Configuration (`llm`)

Setting	Description	Default	Access Level	Environment Variable
provider_enum	LLM provider to use ('ollama' or 'openai').	`ollama`	normal	`LLM_PROVIDER`
model	Default LLM to use for the selected provider.	`llama3.2`	normal	`LLM_MODEL`
models	List of LLMs the user can choose from.	See code default	normal	`LLM_MODELS`

Ollama Configuration (`ollama`)

Setting	Description	Default	Access Level	Environment Variable
ip	IP address for the Ollama API endpoint.	`localhost`	protected	`OLLAMA_IP`
port	Port for the Ollama API endpoint.	`11434`	protected	`OLLAMA_PORT`
num_ctx	Context window size for token generation.	`4096`	normal	`OLLAMA_NUM_CTX`
repeat_last_n	How far back to look to prevent repetition.	`64`	normal	`OLLAMA_REPEAT_LAST_N`
repeat_penalty	Strength of repetition penalty.	`1.1`	normal	`OLLAMA_REPEAT_PENALTY`
temperature	Sampling temperature.	`0.8`	normal	`OLLAMA_TEMPERATURE`
timeout	Timeout in seconds for API requests.	`30.0`	normal	`OLLAMA_TIMEOUT`
seed	Random seed for generation.	`0`	normal	`OLLAMA_SEED`
stop	Stop sequences for generation.	`None`	normal
num_predict	Maximum tokens to predict.	`-1`	normal	`OLLAMA_NUM_PREDICT`
top_k	Top-k sampling parameter.	`40`	normal	`OLLAMA_TOP_K`
top_p	Top-p sampling parameter.	`0.9`	normal	`OLLAMA_TOP_P`
min_p	Minimum probability for sampling.	`0.0`	normal	`OLLAMA_MIN_P`

OpenAI Configuration (`openai`)

Setting	Description	Default	Access Level	Environment Variable
api_key	API key for OpenAI services.	From env	protected	`OPENAI_API_KEY`
api_base	Base URL for OpenAI API requests.	`https://api.openai.com/v1`	read_only
timeout	Timeout in seconds for API requests.	`60`	normal	`OPENAI_TIMEOUT`
max_completion_tokens	Maximum tokens for completions.	`2048`	normal	`OPENAI_MAX_COMPLETION_TOKENS`
temperature	Sampling temperature.	`0.7`	normal	`OPENAI_TEMPERATURE`
top_p	Nucleus sampling parameter.	`1.0`	normal	`OPENAI_TOP_P`
tool_choice	Tool choice for API requests.	`auto`	normal	`OPENAI_TOOL_CHOICE`

Path Configuration (`paths`)

Setting	Description	Default	Access Level	Environment Variable
hatchling_source_dir	Directory where Hatchling source code is located.	Auto-detected	read_only	`HATCHLING_SOURCE_DIR`
envs_dir	Directory for Hatch environments.	Auto-detected	read_only	`HATCH_ENVS_DIR`
hatchling_cache_dir	Directory for Hatchling cache and data storage.	`~/.hatch`	read_only	`HATCHLING_CACHE_DIR`
hatchling_settings_dir	Directory for Hatchling settings storage.	`~/.hatch/settings`	read_only	`HATCHLING_SETTINGS_DIR`

Tool Calling (`tool_calling`)

Setting	Description	Default	Access Level	Environment Variable
max_iterations	Maximum number of tool call iterations.	`5`	normal
max_working_time	Maximum time in seconds for tool operations.	`60.0`	normal
max_tool_working_time	Maximum time in seconds for a single tool operation.	`12.0`	normal

User Interface (`ui`)

Setting	Description	Default	Access Level	Environment Variable
language_code	Language code for user interface localization.	`en`	normal	`HATCHLING_DEFAULT_LANGUAGE`
language_code	Language code for user interface localization.	`en`	`fr`	normal