配置

lazyllm.Config

Config是LazyLLM提供的配置类，可以支持通过加载配置文件、设置环境变量、编码写入默认值等方式设置LazyLLM框架的相关配置，以及导出当前所有的配置项和对应的值。 Config模块自动生成一个config对象，其中包含所有的配置。

Parameters:

• prefix (str, default: 'LAZYLLM' ) –

  环境变量前缀。默认为 'LAZYLLM'

• home (str, default: None ) –

  配置文件目录路径。默认为 '~/.lazyllm'

LazyLLM Configurations:

• home:
  
• Description: The default home directory for LazyLLM., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_HOME
  
  
• mode:
  
• Description: The default mode for LazyLLM., type: Mode, default: {desc["default"]}
  
• Environment Variable:
  
  • LAZYLLM_DISPLAY: Mode.Display
    
  • LAZYLLM_DEBUG: Mode.Debug
    
    
• repr_ml:
  
• Description: Whether to use Markup Language for repr., type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_REPR_USE_ML
  
  
• repr_show_child:
  
• Description: Whether to show child modules in repr., type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_REPR_SHOW_CHILD
  
  
• rag_store:
  
• Description: The default store for RAG., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_RAG_STORE
  
  
• gpu_type:
  
• Description: The default GPU type for LazyLLM., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_GPU_TYPE
  
  
• train_target_root:
  
• Description: The default target root for training., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_TRAIN_TARGET_ROOT
  
  
• infer_log_root:
  
• Description: The default log root for inference., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_INFER_LOG_ROOT
  
  
• temp_dir:
  
• Description: The default temp directory for LazyLLM., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_TEMP_DIR
  
  
• thread_pool_worker_num:
  
• Description: The default number of workers for thread pool., type: int, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_THREAD_POOL_WORKER_NUM
  
  
• deploy_skip_check_kw:
  
• Description: Whether to skip check keywords for deployment., type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_DEPLOY_SKIP_CHECK_KW
  
  
• allow_internal_network:
  
• Description: Whether to allow loading images from internal network addresses. Set to False for security in production environments., type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_ALLOW_INTERNAL_NETWORK
  
  
• redis_url:
  
• Description: The URL of the Redis server., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_REDIS_URL
  
  
• redis_recheck_delay:
  
• Description: The delay of the Redis server check., type: int, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_REDIS_RECHECK_DELAY
  
  
• use_builtin:
  
• Description: Whether to use registry modules in python builtin., type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_USE_BUILTIN
  
  
• default_fsqueue:
  
• Description: The default file system queue to use., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_DEFAULT_FSQUEUE
  
  
• fsqredis_url:
  
• Description: The URL of the Redis server for the file system queue., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_FSQREDIS_URL
  
  
• default_recent_k:
  
• Description: The number of recent inputs that RecentQueue keeps track of., type: int, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_DEFAULT_RECENT_K
  
  
• save_flow_result:
  
• Description: Whether to save the intermediate result of the pipeline., type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_SAVE_FLOW_RESULT
  
  
• parallel_multiprocessing:
  
• Description: Whether to use multiprocessing for parallel execution, if not, default to use threading., type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_PARALLEL_MULTIPROCESSING
  
  
• launcher:
  
• Description: The default remote launcher to use if no launcher is specified., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_DEFAULT_LAUNCHER
  
  
• cuda_visible:
  
• Description: Whether to set the CUDA_VISIBLE_DEVICES environment variable., type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_CUDA_VISIBLE
  
  
• partition:
  
• Description: The default slurm partition to use if no partition is specified., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_SLURM_PART
  
  
• sco.workspace:
  
• Description: The default SCO workspace to use if no workspace is specified., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_SCO_WORKSPACE
  
  
• sco_env_name:
  
• Description: The default SCO environment name to use if no environment name is specified., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_SCO_ENV_NAME
  
  
• sco_keep_record:
  
• Description: Whether to keep the record of the Sensecore job., type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_SCO_KEEP_RECORD
  
  
• sco_resource_type:
  
• Description: The default SCO resource type to use if no resource type is specified., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_SCO_RESOURCE_TYPE
  
  
• k8s_env_name:
  
• Description: The default k8s environment name to use if no environment name is specified., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_K8S_ENV_NAME
  
  
• k8s_config_path:
  
• Description: The default k8s configuration path to use if no configuration path is specified., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_K8S_CONFIG_PATH
  
  
• k8s_device_type:
  
• Description: The default k8s device type to use if no device type is specified., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_K8S_DEVICE_TYPE
  
  
• model_source:
  
• Description: The default model source to use., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_MODEL_SOURCE
  
  
• model_cache_dir:
  
• Description: The default model cache directory to use(Read and Write)., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_MODEL_CACHE_DIR
  
  
• model_path:
  
• Description: The default model path to use(ReadOnly)., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_MODEL_PATH
  
  
• model_source_token:
  
• Description: The default token for configed model source(hf or ms) to use., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_MODEL_SOURCE_TOKEN
  
  
• data_path:
  
• Description: The default data path to use., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_DATA_PATH
  
  
• openai_api:
  
• Description: Whether to use OpenAI API for vllm deployer., type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_OPENAI_API
  
  
• use_ray:
  
• Description: Whether to use Ray for ServerModule(relay server)., type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_USE_RAY
  
  
• num_gpus_per_node:
  
• Description: The number of GPUs per node for Ray launcher when deploy models., type: int, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_NUM_GPUS_PER_NODE
  
  
• lmdeploy_eager_mode:
  
• Description: Whether to use eager mode for lmdeploy., type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_LMDEPLOY_EAGER_MODE
  
  
• default_embedding_engine:
  
• Description: The default embedding engine to use., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_DEFAULT_EMBEDDING_ENGINE
  
  
• mindie_home:
  
• Description: The home directory of MindIE., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_MINDIE_HOME
  
  
• gpu_memory:
  
• Description: The memory of the GPU., type: int, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_GPU_MEMORY
  
  
• cache_dir:
  
• Description: The default result cache directory for module to use(Read and Write)., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_CACHE_DIR
  
  
• cache_strategy:
  
• Description: The default cache strategy to use(memory, file, sqlite, redis)., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_CACHE_STRATEGY
  
  
• cache_mode:
  
• Description: The default cache mode to use(Read and Write, Read Only, Write Only, None)., type: str, default: {desc["default"]}
  
• Options: RW, RO, WO, NONE
  
• Environment Variable: LAZYLLM_CACHE_MODE
  
  
• cache_online_module:
  
• Description: Whether to cache the online module result. Use for unit test., type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_CACHE_ONLINE_MODULE
  
  
• aiping_api_key:
  
• Description: The API key for aiping, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_AIPING_API_KEY
  
  
• aiping_model_name:
  
• Description: The default model name for aiping, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_AIPING_MODEL_NAME
  
  
• aiping_text2image_model_name:
  
• Description: The default text2image model name for aiping, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_AIPING_TEXT2IMAGE_MODEL_NAME
  
  
• deepseek_api_key:
  
• Description: The API key for deepseek, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_DEEPSEEK_API_KEY
  
  
• deepseek_model_name:
  
• Description: The default model name for deepseek, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_DEEPSEEK_MODEL_NAME
  
  
• doubao_api_key:
  
• Description: The API key for doubao, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_DOUBAO_API_KEY
  
  
• doubao_model_name:
  
• Description: The default model name for doubao, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_DOUBAO_MODEL_NAME
  
  
• doubao_text2image_model_name:
  
• Description: The default text2image model name for doubao, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_DOUBAO_TEXT2IMAGE_MODEL_NAME
  
  
• glm_api_key:
  
• Description: The API key for glm, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_GLM_API_KEY
  
  
• glm_model_name:
  
• Description: The default model name for glm, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_GLM_MODEL_NAME
  
  
• glm_stt_model_name:
  
• Description: The default stt model name for glm, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_GLM_STT_MODEL_NAME
  
  
• glm_text2image_model_name:
  
• Description: The default text2image model name for glm, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_GLM_TEXT2IMAGE_MODEL_NAME
  
  
• kimi_api_key:
  
• Description: The API key for kimi, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_KIMI_API_KEY
  
  
• kimi_model_name:
  
• Description: The default model name for kimi, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_KIMI_MODEL_NAME
  
  
• minimax_api_key:
  
• Description: The API key for minimax, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_MINIMAX_API_KEY
  
  
• minimax_model_name:
  
• Description: The default model name for minimax, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_MINIMAX_MODEL_NAME
  
  
• minimax_text2image_model_name:
  
• Description: The default text2image model name for minimax, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_MINIMAX_TEXT2IMAGE_MODEL_NAME
  
  
• minimax_tts_model_name:
  
• Description: The default tts model name for minimax, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_MINIMAX_TTS_MODEL_NAME
  
  
• openai_api_key:
  
• Description: The API key for openai, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_OPENAI_API_KEY
  
  
• openai_model_name:
  
• Description: The default model name for openai, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_OPENAI_MODEL_NAME
  
  
• ppio_api_key:
  
• Description: The API key for ppio, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_PPIO_API_KEY
  
  
• ppio_model_name:
  
• Description: The default model name for ppio, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_PPIO_MODEL_NAME
  
  
• qwen_api_key:
  
• Description: The API key for qwen, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_QWEN_API_KEY
  
  
• qwen_model_name:
  
• Description: The default model name for qwen, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_QWEN_MODEL_NAME
  
  
• qwen_stt_model_name:
  
• Description: The default stt model name for qwen, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_QWEN_STT_MODEL_NAME
  
  
• qwen_text2image_model_name:
  
• Description: The default text2image model name for qwen, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_QWEN_TEXT2IMAGE_MODEL_NAME
  
  
• qwen_tts_model_name:
  
• Description: The default tts model name for qwen, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_QWEN_TTS_MODEL_NAME
  
  
• sensenova_secret_key:
  
• Description: The secret key for SenseNova., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_SENSENOVA_SECRET_KEY
  
  
• sensenova_api_key:
  
• Description: The API key for sensenova, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_SENSENOVA_API_KEY
  
  
• sensenova_model_name:
  
• Description: The default model name for sensenova, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_SENSENOVA_MODEL_NAME
  
  
• siliconflow_api_key:
  
• Description: The API key for siliconflow, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_SILICONFLOW_API_KEY
  
  
• siliconflow_model_name:
  
• Description: The default model name for siliconflow, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_SILICONFLOW_MODEL_NAME
  
  
• siliconflow_text2image_model_name:
  
• Description: The default text2image model name for siliconflow, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_SILICONFLOW_TEXT2IMAGE_MODEL_NAME
  
  
• siliconflow_tts_model_name:
  
• Description: The default tts model name for siliconflow, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_SILICONFLOW_TTS_MODEL_NAME
  
  
• trainable_module_config_map_path:
  
• Description: The default path for trainable module config map., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_TRAINABLE_MODULE_CONFIG_MAP_PATH
  
  
• auto_model_config_map_path:
  
• Description: The default path for automodel config map., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_AUTO_MODEL_CONFIG_MAP_PATH
  
  
• trainable_magic_mock:
  
• Description: Whether to use magic mock for trainable module(used for unit test)., type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_TRAINABLE_MAGIC_MOCK
  
  
• cache_local_module:
  
• Description: Whether to cache the local module result. Use for unit test., type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_CACHE_LOCAL_MODULE
  
  
• raise_on_add_doc_error:
  
• Description: Whether to raise an error when adding doc failed., type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_RAISE_ON_ADD_DOC_ERROR
  
  
• language:
  
• Description: The language of the documentation., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_LANGUAGE
  
  
• init_doc:
  
• Description: whether to init docs, type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_INIT_DOC
  
  
• max_embedding_workers:
  
• Description: The default number of workers for embedding in RAG., type: int, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_MAX_EMBEDDING_WORKERS
  
  
• default_dlmanager:
  
• Description: The default document list manager for RAG., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_DEFAULT_DOCLIST_MANAGER
  
  
• auto_detect_encoding:
  
• Description: Whether auto detecting txt encoding, type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_AUTO_DETECT_ENCODING
  
  
• enable_chardet:
  
• Description: Whether to use chardet when detect txt encoding, type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_ENABLE_CHARDET
  
  
• use_encoding_cache:
  
• Description: Whether use cahce to accelerate txt encoding, type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_USE_ENCODING_CACHE
  
  
• paddleocr_api_key:
  
• Description: The API key for PaddleOCR, type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_PADDLEOCR_API_KEY
  
  
• rag_filename_as_id:
  
• Description: Whether to use filename as id for RAG., type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_RAG_FILENAME_AS_ID
  
  
• use_fallback_reader:
  
• Description: Whether to use fallback reader for RAG., type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_USE_FALLBACK_READER
  
  
• eval_result_dir:
  
• Description: The default result directory for eval., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_EVAL_RESULT_DIR
  
  
• debug:
  
• Description: Whether to enable debug mode., type: bool, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_DEBUG
  
  
• log_name:
  
• Description: The name of the log file., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_LOG_NAME
  
  
• expected_log_modules:
  
• Description: The expected log modules, separated by comma., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_EXPECTED_LOG_MODULES
  
  
• log_level:
  
• Description: The level of the log., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_LOG_LEVEL
  
  
• log_format:
  
• Description: The format of the log., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_LOG_FORMAT
  
  
• log_dir:
  
• Description: The directory of the log file., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_LOG_DIR
  
  
• log_file_level:
  
• Description: The level of the log file., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_LOG_FILE_LEVEL
  
  
• log_file_size:
  
• Description: The size of the log file., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_LOG_FILE_SIZE
  
  
• log_file_retention:
  
• Description: The retention of the log file., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_LOG_FILE_RETENTION
  
  
• log_file_mode:
  
• Description: The mode of the log file., type: str, default: {desc["default"]}
  
• Environment Variable: LAZYLLM_LOG_FILE_MODE
  

Source code in lazyllm/configs.py

class Config(metaclass=_ConfigMeta):
    """Config是LazyLLM提供的配置类，可以支持通过加载配置文件、设置环境变量、编码写入默认值等方式设置LazyLLM框架的相关配置，以及导出当前所有的配置项和对应的值。
Config模块自动生成一个config对象，其中包含所有的配置。

Args:
    prefix (str, optional): 环境变量前缀。默认为 'LAZYLLM'
    home (str, optional): 配置文件目录路径。默认为 '~/.lazyllm'

**LazyLLM Configurations:**

- **home**:<br>
  - Description: The default home directory for LazyLLM., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_HOME<br>
<br>
- **mode**:<br>
  - Description: The default mode for LazyLLM., type: `Mode`, default: `{desc["default"]}`<br>
  - Environment Variable:<br>
    - LAZYLLM_DISPLAY: Mode.Display<br>
    - LAZYLLM_DEBUG: Mode.Debug<br>
<br>
- **repr_ml**:<br>
  - Description: Whether to use Markup Language for repr., type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_REPR_USE_ML<br>
<br>
- **repr_show_child**:<br>
  - Description: Whether to show child modules in repr., type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_REPR_SHOW_CHILD<br>
<br>
- **rag_store**:<br>
  - Description: The default store for RAG., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_RAG_STORE<br>
<br>
- **gpu_type**:<br>
  - Description: The default GPU type for LazyLLM., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_GPU_TYPE<br>
<br>
- **train_target_root**:<br>
  - Description: The default target root for training., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_TRAIN_TARGET_ROOT<br>
<br>
- **infer_log_root**:<br>
  - Description: The default log root for inference., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_INFER_LOG_ROOT<br>
<br>
- **temp_dir**:<br>
  - Description: The default temp directory for LazyLLM., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_TEMP_DIR<br>
<br>
- **thread_pool_worker_num**:<br>
  - Description: The default number of workers for thread pool., type: `int`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_THREAD_POOL_WORKER_NUM<br>
<br>
- **deploy_skip_check_kw**:<br>
  - Description: Whether to skip check keywords for deployment., type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_DEPLOY_SKIP_CHECK_KW<br>
<br>
- **allow_internal_network**:<br>
  - Description: Whether to allow loading images from internal network addresses. Set to False for security in production environments., type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_ALLOW_INTERNAL_NETWORK<br>
<br>
- **redis_url**:<br>
  - Description: The URL of the Redis server., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_REDIS_URL<br>
<br>
- **redis_recheck_delay**:<br>
  - Description: The delay of the Redis server check., type: `int`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_REDIS_RECHECK_DELAY<br>
<br>
- **use_builtin**:<br>
  - Description: Whether to use registry modules in python builtin., type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_USE_BUILTIN<br>
<br>
- **default_fsqueue**:<br>
  - Description: The default file system queue to use., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_DEFAULT_FSQUEUE<br>
<br>
- **fsqredis_url**:<br>
  - Description: The URL of the Redis server for the file system queue., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_FSQREDIS_URL<br>
<br>
- **default_recent_k**:<br>
  - Description: The number of recent inputs that RecentQueue keeps track of., type: `int`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_DEFAULT_RECENT_K<br>
<br>
- **save_flow_result**:<br>
  - Description: Whether to save the intermediate result of the pipeline., type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_SAVE_FLOW_RESULT<br>
<br>
- **parallel_multiprocessing**:<br>
  - Description: Whether to use multiprocessing for parallel execution, if not, default to use threading., type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_PARALLEL_MULTIPROCESSING<br>
<br>
- **launcher**:<br>
  - Description: The default remote launcher to use if no launcher is specified., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_DEFAULT_LAUNCHER<br>
<br>
- **cuda_visible**:<br>
  - Description: Whether to set the CUDA_VISIBLE_DEVICES environment variable., type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_CUDA_VISIBLE<br>
<br>
- **partition**:<br>
  - Description: The default slurm partition to use if no partition is specified., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_SLURM_PART<br>
<br>
- **sco.workspace**:<br>
  - Description: The default SCO workspace to use if no workspace is specified., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_SCO_WORKSPACE<br>
<br>
- **sco_env_name**:<br>
  - Description: The default SCO environment name to use if no environment name is specified., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_SCO_ENV_NAME<br>
<br>
- **sco_keep_record**:<br>
  - Description: Whether to keep the record of the Sensecore job., type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_SCO_KEEP_RECORD<br>
<br>
- **sco_resource_type**:<br>
  - Description: The default SCO resource type to use if no resource type is specified., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_SCO_RESOURCE_TYPE<br>
<br>
- **k8s_env_name**:<br>
  - Description: The default k8s environment name to use if no environment name is specified., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_K8S_ENV_NAME<br>
<br>
- **k8s_config_path**:<br>
  - Description: The default k8s configuration path to use if no configuration path is specified., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_K8S_CONFIG_PATH<br>
<br>
- **k8s_device_type**:<br>
  - Description: The default k8s device type to use if no device type is specified., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_K8S_DEVICE_TYPE<br>
<br>
- **model_source**:<br>
  - Description: The default model source to use., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_MODEL_SOURCE<br>
<br>
- **model_cache_dir**:<br>
  - Description: The default model cache directory to use(Read and Write)., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_MODEL_CACHE_DIR<br>
<br>
- **model_path**:<br>
  - Description: The default model path to use(ReadOnly)., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_MODEL_PATH<br>
<br>
- **model_source_token**:<br>
  - Description: The default token for configed model source(hf or ms) to use., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_MODEL_SOURCE_TOKEN<br>
<br>
- **data_path**:<br>
  - Description: The default data path to use., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_DATA_PATH<br>
<br>
- **openai_api**:<br>
  - Description: Whether to use OpenAI API for vllm deployer., type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_OPENAI_API<br>
<br>
- **use_ray**:<br>
  - Description: Whether to use Ray for ServerModule(relay server)., type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_USE_RAY<br>
<br>
- **num_gpus_per_node**:<br>
  - Description: The number of GPUs per node for Ray launcher when deploy models., type: `int`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_NUM_GPUS_PER_NODE<br>
<br>
- **lmdeploy_eager_mode**:<br>
  - Description: Whether to use eager mode for lmdeploy., type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_LMDEPLOY_EAGER_MODE<br>
<br>
- **default_embedding_engine**:<br>
  - Description: The default embedding engine to use., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_DEFAULT_EMBEDDING_ENGINE<br>
<br>
- **mindie_home**:<br>
  - Description: The home directory of MindIE., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_MINDIE_HOME<br>
<br>
- **gpu_memory**:<br>
  - Description: The memory of the GPU., type: `int`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_GPU_MEMORY<br>
<br>
- **cache_dir**:<br>
  - Description: The default result cache directory for module to use(Read and Write)., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_CACHE_DIR<br>
<br>
- **cache_strategy**:<br>
  - Description: The default cache strategy to use(memory, file, sqlite, redis)., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_CACHE_STRATEGY<br>
<br>
- **cache_mode**:<br>
  - Description: The default cache mode to use(Read and Write, Read Only, Write Only, None)., type: `str`, default: `{desc["default"]}`<br>
  - Options: RW, RO, WO, NONE<br>
  - Environment Variable: LAZYLLM_CACHE_MODE<br>
<br>
- **cache_online_module**:<br>
  - Description: Whether to cache the online module result. Use for unit test., type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_CACHE_ONLINE_MODULE<br>
<br>
- **aiping_api_key**:<br>
  - Description: The API key for aiping, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_AIPING_API_KEY<br>
<br>
- **aiping_model_name**:<br>
  - Description: The default model name for aiping, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_AIPING_MODEL_NAME<br>
<br>
- **aiping_text2image_model_name**:<br>
  - Description: The default text2image model name for aiping, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_AIPING_TEXT2IMAGE_MODEL_NAME<br>
<br>
- **deepseek_api_key**:<br>
  - Description: The API key for deepseek, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_DEEPSEEK_API_KEY<br>
<br>
- **deepseek_model_name**:<br>
  - Description: The default model name for deepseek, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_DEEPSEEK_MODEL_NAME<br>
<br>
- **doubao_api_key**:<br>
  - Description: The API key for doubao, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_DOUBAO_API_KEY<br>
<br>
- **doubao_model_name**:<br>
  - Description: The default model name for doubao, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_DOUBAO_MODEL_NAME<br>
<br>
- **doubao_text2image_model_name**:<br>
  - Description: The default text2image model name for doubao, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_DOUBAO_TEXT2IMAGE_MODEL_NAME<br>
<br>
- **glm_api_key**:<br>
  - Description: The API key for glm, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_GLM_API_KEY<br>
<br>
- **glm_model_name**:<br>
  - Description: The default model name for glm, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_GLM_MODEL_NAME<br>
<br>
- **glm_stt_model_name**:<br>
  - Description: The default stt model name for glm, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_GLM_STT_MODEL_NAME<br>
<br>
- **glm_text2image_model_name**:<br>
  - Description: The default text2image model name for glm, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_GLM_TEXT2IMAGE_MODEL_NAME<br>
<br>
- **kimi_api_key**:<br>
  - Description: The API key for kimi, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_KIMI_API_KEY<br>
<br>
- **kimi_model_name**:<br>
  - Description: The default model name for kimi, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_KIMI_MODEL_NAME<br>
<br>
- **minimax_api_key**:<br>
  - Description: The API key for minimax, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_MINIMAX_API_KEY<br>
<br>
- **minimax_model_name**:<br>
  - Description: The default model name for minimax, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_MINIMAX_MODEL_NAME<br>
<br>
- **minimax_text2image_model_name**:<br>
  - Description: The default text2image model name for minimax, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_MINIMAX_TEXT2IMAGE_MODEL_NAME<br>
<br>
- **minimax_tts_model_name**:<br>
  - Description: The default tts model name for minimax, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_MINIMAX_TTS_MODEL_NAME<br>
<br>
- **openai_api_key**:<br>
  - Description: The API key for openai, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_OPENAI_API_KEY<br>
<br>
- **openai_model_name**:<br>
  - Description: The default model name for openai, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_OPENAI_MODEL_NAME<br>
<br>
- **ppio_api_key**:<br>
  - Description: The API key for ppio, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_PPIO_API_KEY<br>
<br>
- **ppio_model_name**:<br>
  - Description: The default model name for ppio, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_PPIO_MODEL_NAME<br>
<br>
- **qwen_api_key**:<br>
  - Description: The API key for qwen, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_QWEN_API_KEY<br>
<br>
- **qwen_model_name**:<br>
  - Description: The default model name for qwen, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_QWEN_MODEL_NAME<br>
<br>
- **qwen_stt_model_name**:<br>
  - Description: The default stt model name for qwen, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_QWEN_STT_MODEL_NAME<br>
<br>
- **qwen_text2image_model_name**:<br>
  - Description: The default text2image model name for qwen, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_QWEN_TEXT2IMAGE_MODEL_NAME<br>
<br>
- **qwen_tts_model_name**:<br>
  - Description: The default tts model name for qwen, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_QWEN_TTS_MODEL_NAME<br>
<br>
- **sensenova_secret_key**:<br>
  - Description: The secret key for SenseNova., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_SENSENOVA_SECRET_KEY<br>
<br>
- **sensenova_api_key**:<br>
  - Description: The API key for sensenova, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_SENSENOVA_API_KEY<br>
<br>
- **sensenova_model_name**:<br>
  - Description: The default model name for sensenova, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_SENSENOVA_MODEL_NAME<br>
<br>
- **siliconflow_api_key**:<br>
  - Description: The API key for siliconflow, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_SILICONFLOW_API_KEY<br>
<br>
- **siliconflow_model_name**:<br>
  - Description: The default model name for siliconflow, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_SILICONFLOW_MODEL_NAME<br>
<br>
- **siliconflow_text2image_model_name**:<br>
  - Description: The default text2image model name for siliconflow, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_SILICONFLOW_TEXT2IMAGE_MODEL_NAME<br>
<br>
- **siliconflow_tts_model_name**:<br>
  - Description: The default tts model name for siliconflow, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_SILICONFLOW_TTS_MODEL_NAME<br>
<br>
- **trainable_module_config_map_path**:<br>
  - Description: The default path for trainable module config map., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_TRAINABLE_MODULE_CONFIG_MAP_PATH<br>
<br>
- **auto_model_config_map_path**:<br>
  - Description: The default path for automodel config map., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_AUTO_MODEL_CONFIG_MAP_PATH<br>
<br>
- **trainable_magic_mock**:<br>
  - Description: Whether to use magic mock for trainable module(used for unit test)., type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_TRAINABLE_MAGIC_MOCK<br>
<br>
- **cache_local_module**:<br>
  - Description: Whether to cache the local module result. Use for unit test., type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_CACHE_LOCAL_MODULE<br>
<br>
- **raise_on_add_doc_error**:<br>
  - Description: Whether to raise an error when adding doc failed., type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_RAISE_ON_ADD_DOC_ERROR<br>
<br>
- **language**:<br>
  - Description: The language of the documentation., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_LANGUAGE<br>
<br>
- **init_doc**:<br>
  - Description: whether to init docs, type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_INIT_DOC<br>
<br>
- **max_embedding_workers**:<br>
  - Description: The default number of workers for embedding in RAG., type: `int`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_MAX_EMBEDDING_WORKERS<br>
<br>
- **default_dlmanager**:<br>
  - Description: The default document list manager for RAG., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_DEFAULT_DOCLIST_MANAGER<br>
<br>
- **auto_detect_encoding**:<br>
  - Description: Whether auto detecting txt encoding, type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_AUTO_DETECT_ENCODING<br>
<br>
- **enable_chardet**:<br>
  - Description: Whether to use chardet when detect txt encoding, type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_ENABLE_CHARDET<br>
<br>
- **use_encoding_cache**:<br>
  - Description: Whether use cahce to accelerate txt encoding, type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_USE_ENCODING_CACHE<br>
<br>
- **paddleocr_api_key**:<br>
  - Description: The API key for PaddleOCR, type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_PADDLEOCR_API_KEY<br>
<br>
- **rag_filename_as_id**:<br>
  - Description: Whether to use filename as id for RAG., type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_RAG_FILENAME_AS_ID<br>
<br>
- **use_fallback_reader**:<br>
  - Description: Whether to use fallback reader for RAG., type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_USE_FALLBACK_READER<br>
<br>
- **eval_result_dir**:<br>
  - Description: The default result directory for eval., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_EVAL_RESULT_DIR<br>
<br>
- **debug**:<br>
  - Description: Whether to enable debug mode., type: `bool`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_DEBUG<br>
<br>
- **log_name**:<br>
  - Description: The name of the log file., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_LOG_NAME<br>
<br>
- **expected_log_modules**:<br>
  - Description: The expected log modules, separated by comma., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_EXPECTED_LOG_MODULES<br>
<br>
- **log_level**:<br>
  - Description: The level of the log., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_LOG_LEVEL<br>
<br>
- **log_format**:<br>
  - Description: The format of the log., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_LOG_FORMAT<br>
<br>
- **log_dir**:<br>
  - Description: The directory of the log file., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_LOG_DIR<br>
<br>
- **log_file_level**:<br>
  - Description: The level of the log file., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_LOG_FILE_LEVEL<br>
<br>
- **log_file_size**:<br>
  - Description: The size of the log file., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_LOG_FILE_SIZE<br>
<br>
- **log_file_retention**:<br>
  - Description: The retention of the log file., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_LOG_FILE_RETENTION<br>
<br>
- **log_file_mode**:<br>
  - Description: The mode of the log file., type: `str`, default: `{desc["default"]}`<br>
  - Environment Variable: LAZYLLM_LOG_FILE_MODE<br>
"""
    def __init__(self, prefix: str = 'LAZYLLM', home: Optional[str] = None, config_file: str = 'config.json'):
        self._prefix = prefix.upper()
        self._impl, self._cfgs = dict(), dict()
        if not home:
            home = '.lazyllm' if self._prefix == 'LAZYLLM' else f'.lazyllm_{prefix.lower()}'
            home = os.path.join(os.path.expanduser('~'), home)
        _ConfigMeta._homes[self._prefix] = home
        self._update_impl('home', str, None, 'HOME')
        os.makedirs(self['home'], exist_ok=True)
        self._cgf_path = os.path.join(self['home'], config_file)
        if os.path.exists(self._cgf_path):
            with open(self._cgf_path, 'r+') as f:
                self._cfgs = Config.get_config(json.loads(f))
        for name, cfg in _ConfigMeta._registered_cfgs.items():
            if name == 'home': continue
            self._update_impl(name, cfg['type'], cfg['default'], cfg['env'])

    def add(self, name: str, type: type, default: Optional[Union[int, str, bool]] = None, env: Union[str, dict] = None,
            *, options: Optional[List] = None, description: Optional[str] = None):
        """将值加载至LazyLLM的配置项中。函数首先尝试从加载自config.json的字典中查找名字为name的值。若找到则从该字典中删去名为name的键。并将对应的值写入config。
若env是一个字符串，函数会调用getenv寻找env对应的LazyLLM环境变量，若找到则写入config。如果env为一个字典，函数将尝试调用getenv寻找字典中的key对于的环境变量并转换为bool型。
若转换完成的bool值是True，则将字典中当前的key对应的值写入config。

Args:
    name (str): 配置项名称
    type (type): 该配置的类型
    default (可选): 若无法获取到任何值，该配置的默认值
    env (可选): 不包含前缀的环境变量名称，或者一个字典，其中的key是不包含前缀的环境变量名称，value是要加入配置的值。
"""
        if (name := name.lower()) in Config.__dict__.keys():
            raise RuntimeError(f'{name} is attribute of Config, please change it!')
        _ConfigMeta.add(name, type, default, env, options=options, description=description)

    def getenv(self, name, type, default=None):
        """用于检查config.json配置文件中是否还有没有通过add方法载入的配置项
Config.getenv(name, type, default): -> str

获取LazyLLM相关环境变量的值

Args:
    name (str): 不包含前缀的环境变量名字，不区分大小写。函数将从拼接了前缀和此名字的全大写的环境变量中获取对应的值。
    type (type): 指定该配置的类型，例如str。对于bool型，函数会将'TRUE', 'True', 1, 'ON', '1'这5种输入转换为True。
    default (可选): 若无法获取到环境变量的值，将返回此变量。
"""
        r = os.getenv(f'{self._prefix}_{name.upper()}', default)
        if type == bool:
            return r in (True, 'TRUE', 'True', 1, 'ON', '1')
        return type(r) if r is not None else r

    @staticmethod
    def get_config(cfg):
        """\ 
静态方法：从配置字典中获取配置。
这是一个简单的配置获取方法，主要用于从已加载的配置字典中提取配置信息。

Args:
    cfg (dict): 从配置文件中读取的配置字典。
"""
        return cfg

    def get_all_configs(self):
        """获取config中所有的配置


Examples:
    >>> import lazyllm
    >>> from lazyllm.configs import config
    >>> config['launcher']
    'empty'
    >>> config.get_all_configs()
    {'home': '~/.lazyllm/', 'mode': <Mode.Normal: (1,)>, 'repr_ml': False, 'rag_store': 'None', 'redis_url': 'None', ...}
    """
        return self._impl

    def done(self):
        """用于检查config.json配置文件中是否还有没有通过add方法载入的配置项
"""
        ins = _ConfigMeta._instances['lazyllm']
        assert len(ins._cfgs) == 0, f'Invalid cfgs ({"".join(ins._cfgs.keys())}) are given in {ins._cgf_path}'

    @contextmanager
    def temp(self, name, value):
        """
临时修改配置项的上下文管理器。在with语句块内临时修改指定配置项的值，退出语句块后自动恢复原值。注意，此函数并非线程安全，请勿在多线程或者多协程的环境下使用。

Args:
    name (str): 要临时修改的配置项名称。
    value (Any): 临时设置的值。
"""
        old_value = self[name]
        self._impl[name] = value
        yield
        self._impl[name] = old_value

    def _update_impl(self, name: str, type: type, default: Optional[Union[int, str, bool]] = None,
                     env: Union[str, dict, list] = None):
        self._impl[name] = self._cfgs.pop(name) if name in self._cfgs else (
            _ConfigMeta._get_default_home(self._prefix) if name == 'home' else default)
        if isinstance(env, dict):
            for k, v in env.items():
                if self.getenv(k, bool):
                    self._impl[name] = v
                    break
        elif isinstance(env, list):
            for k in env:
                if (v := self.getenv(k, type)):
                    self._impl[name] = v
                    break
        elif env:
            self._impl[name] = self.getenv(env, type, self._impl[name])
        if not isinstance(self._impl[name], type) and self._impl[name] is not None: raise TypeError(
            f'Invalid config type for {name}, type is {type}')

    def __getitem__(self, name):
        try:
            if isinstance(name, bytes): name = name.decode('utf-8')
            name = name.lower()
            if name.startswith(f'{self._prefix.lower()}_'): name = name[len(self._prefix) + 1:]
            return self._impl[name]
        except KeyError as e:
            raise KeyError(f'Error occured when getting key `{name}` from lazyllm global config, msg is: {e}')

    def __str__(self):
        return str(self._impl)

    @property
    def _envs(self):
        return [f'{self._prefix}_{e}'.lower() for e in _ConfigMeta._env_name_map.keys()]

    def refresh(self, targets: Union[bytes, str, List[str]] = None) -> None:
        """
根据环境变量的最新值刷新配置项。如果传入 targets 为字符串，则按单个配置项更新；如果为列表，则批量更新；如果为 None，则扫描所有已映射到环境变量的配置项并更新。

Args:
    targets (str | list[str] | None): 要刷新的配置项名称或列表，传 None 表示刷新所有可从环境变量读取的项。
"""
        names, all_envs = targets, self._envs
        if isinstance(targets, bytes): targets = targets.decode('utf-8')
        if isinstance(targets, str): names = [targets.lower()]
        elif targets is None:
            names = [key.lower() for key in os.environ.keys() if key.lower() in all_envs]
        assert isinstance(names, list)
        for name in names:
            if name.lower() in all_envs:
                name = _ConfigMeta._env_name_map[name[len(self._prefix) + 1:]]
            elif name in Config: continue
            cfg = _ConfigMeta._registered_cfgs[name]
            if name in self._impl: self._update_impl(name, cfg['type'], cfg['default'], cfg['env'])

done()

用于检查config.json配置文件中是否还有没有通过add方法载入的配置项

Source code in lazyllm/configs.py

    def done(self):
        """用于检查config.json配置文件中是否还有没有通过add方法载入的配置项
"""
        ins = _ConfigMeta._instances['lazyllm']
        assert len(ins._cfgs) == 0, f'Invalid cfgs ({"".join(ins._cfgs.keys())}) are given in {ins._cgf_path}'

getenv(name, type, default=None)

用于检查config.json配置文件中是否还有没有通过add方法载入的配置项 Config.getenv(name, type, default): -> str

获取LazyLLM相关环境变量的值

Parameters:

• name (str) –

  不包含前缀的环境变量名字，不区分大小写。函数将从拼接了前缀和此名字的全大写的环境变量中获取对应的值。

• type (type) –

  指定该配置的类型，例如str。对于bool型，函数会将'TRUE', 'True', 1, 'ON', '1'这5种输入转换为True。

• default (可选, default: None ) –

  若无法获取到环境变量的值，将返回此变量。

Source code in lazyllm/configs.py

    def getenv(self, name, type, default=None):
        """用于检查config.json配置文件中是否还有没有通过add方法载入的配置项
Config.getenv(name, type, default): -> str

获取LazyLLM相关环境变量的值

Args:
    name (str): 不包含前缀的环境变量名字，不区分大小写。函数将从拼接了前缀和此名字的全大写的环境变量中获取对应的值。
    type (type): 指定该配置的类型，例如str。对于bool型，函数会将'TRUE', 'True', 1, 'ON', '1'这5种输入转换为True。
    default (可选): 若无法获取到环境变量的值，将返回此变量。
"""
        r = os.getenv(f'{self._prefix}_{name.upper()}', default)
        if type == bool:
            return r in (True, 'TRUE', 'True', 1, 'ON', '1')
        return type(r) if r is not None else r

add(name, type, default=None, env=None, *, options=None, description=None)

将值加载至LazyLLM的配置项中。函数首先尝试从加载自config.json的字典中查找名字为name的值。若找到则从该字典中删去名为name的键。并将对应的值写入config。 若env是一个字符串，函数会调用getenv寻找env对应的LazyLLM环境变量，若找到则写入config。如果env为一个字典，函数将尝试调用getenv寻找字典中的key对于的环境变量并转换为bool型。 若转换完成的bool值是True，则将字典中当前的key对应的值写入config。

Parameters:

• name (str) –

  配置项名称

• type (type) –

  该配置的类型

• default (可选, default: None ) –

  若无法获取到任何值，该配置的默认值

• env (可选, default: None ) –

  不包含前缀的环境变量名称，或者一个字典，其中的key是不包含前缀的环境变量名称，value是要加入配置的值。

Source code in lazyllm/configs.py

    def add(self, name: str, type: type, default: Optional[Union[int, str, bool]] = None, env: Union[str, dict] = None,
            *, options: Optional[List] = None, description: Optional[str] = None):
        """将值加载至LazyLLM的配置项中。函数首先尝试从加载自config.json的字典中查找名字为name的值。若找到则从该字典中删去名为name的键。并将对应的值写入config。
若env是一个字符串，函数会调用getenv寻找env对应的LazyLLM环境变量，若找到则写入config。如果env为一个字典，函数将尝试调用getenv寻找字典中的key对于的环境变量并转换为bool型。
若转换完成的bool值是True，则将字典中当前的key对应的值写入config。

Args:
    name (str): 配置项名称
    type (type): 该配置的类型
    default (可选): 若无法获取到任何值，该配置的默认值
    env (可选): 不包含前缀的环境变量名称，或者一个字典，其中的key是不包含前缀的环境变量名称，value是要加入配置的值。
"""
        if (name := name.lower()) in Config.__dict__.keys():
            raise RuntimeError(f'{name} is attribute of Config, please change it!')
        _ConfigMeta.add(name, type, default, env, options=options, description=description)

get_all_configs()

获取config中所有的配置

Examples:

>>> import lazyllm
>>> from lazyllm.configs import config
>>> config['launcher']
'empty'
>>> config.get_all_configs()
{'home': '~/.lazyllm/', 'mode': <Mode.Normal: (1,)>, 'repr_ml': False, 'rag_store': 'None', 'redis_url': 'None', ...}

Source code in lazyllm/configs.py

    def get_all_configs(self):
        """获取config中所有的配置


Examples:
    >>> import lazyllm
    >>> from lazyllm.configs import config
    >>> config['launcher']
    'empty'
    >>> config.get_all_configs()
    {'home': '~/.lazyllm/', 'mode': <Mode.Normal: (1,)>, 'repr_ml': False, 'rag_store': 'None', 'redis_url': 'None', ...}
    """
        return self._impl

get_config(cfg) staticmethod

\ 静态方法：从配置字典中获取配置。 这是一个简单的配置获取方法，主要用于从已加载的配置字典中提取配置信息。

Parameters:

• cfg (dict) –

  从配置文件中读取的配置字典。

Source code in lazyllm/configs.py

    @staticmethod
    def get_config(cfg):
        """\ 
静态方法：从配置字典中获取配置。
这是一个简单的配置获取方法，主要用于从已加载的配置字典中提取配置信息。

Args:
    cfg (dict): 从配置文件中读取的配置字典。
"""
        return cfg

temp(name, value)

临时修改配置项的上下文管理器。在with语句块内临时修改指定配置项的值，退出语句块后自动恢复原值。注意，此函数并非线程安全，请勿在多线程或者多协程的环境下使用。

Parameters:

• name (str) –

  要临时修改的配置项名称。

• value (Any) –

  临时设置的值。

Source code in lazyllm/configs.py

    @contextmanager
    def temp(self, name, value):
        """
临时修改配置项的上下文管理器。在with语句块内临时修改指定配置项的值，退出语句块后自动恢复原值。注意，此函数并非线程安全，请勿在多线程或者多协程的环境下使用。

Args:
    name (str): 要临时修改的配置项名称。
    value (Any): 临时设置的值。
"""
        old_value = self[name]
        self._impl[name] = value
        yield
        self._impl[name] = old_value

refresh(targets=None)

根据环境变量的最新值刷新配置项。如果传入 targets 为字符串，则按单个配置项更新；如果为列表，则批量更新；如果为 None，则扫描所有已映射到环境变量的配置项并更新。

Parameters:

• targets (str | list[str] | None, default: None ) –

  要刷新的配置项名称或列表，传 None 表示刷新所有可从环境变量读取的项。

Source code in lazyllm/configs.py

    def refresh(self, targets: Union[bytes, str, List[str]] = None) -> None:
        """
根据环境变量的最新值刷新配置项。如果传入 targets 为字符串，则按单个配置项更新；如果为列表，则批量更新；如果为 None，则扫描所有已映射到环境变量的配置项并更新。

Args:
    targets (str | list[str] | None): 要刷新的配置项名称或列表，传 None 表示刷新所有可从环境变量读取的项。
"""
        names, all_envs = targets, self._envs
        if isinstance(targets, bytes): targets = targets.decode('utf-8')
        if isinstance(targets, str): names = [targets.lower()]
        elif targets is None:
            names = [key.lower() for key in os.environ.keys() if key.lower() in all_envs]
        assert isinstance(names, list)
        for name in names:
            if name.lower() in all_envs:
                name = _ConfigMeta._env_name_map[name[len(self._prefix) + 1:]]
            elif name in Config: continue
            cfg = _ConfigMeta._registered_cfgs[name]
            if name in self._impl: self._update_impl(name, cfg['type'], cfg['default'], cfg['env'])