mindformers.AutoConfig
- class mindformers.AutoConfig[source]
This is a generic configuration class that will be instantiated as one of the configuration classes of the library when created with the
from_pretrained()class method. This class cannot be instantiated directly using __init__() (throws an error).Examples
>>> from mindformers import AutoConfig >>> # 1) instantiates a config by yaml model name >>> config_a = AutoConfig.from_pretrained('configs/llama2/predict_llama2_7b.yaml') >>> # 2) instantiates a config by json file >>> config_b = AutoConfig.from_pretrained('./config.json') >>> # 3) instantiates a config by directory containing a configuration json file >>> config_c = AutoConfig.from_pretrained('./dir/') >>> # 4) instantiates a config by model_id from modelers.cn >>> config_d = AutoConfig.from_pretrained('MindSpore-Lab/glm2_6b')
- classmethod from_pretrained(yaml_name_or_path, **kwargs)[source]
From pretrain method, which instantiates a config by YAML, JSON, directory or model_id from modelers.cn.
Warning
The API is experimental and may have some slight breaking changes in the next releases.
- Parameters
yaml_name_or_path (str) – YAML file path, JSON file path, a folder containing a config.json file, or a model_id from modelers.cn. The last three are experimental features.
kwargs (Dict[str, Any], optional) – The values in kwargs of any keys which are configuration attributes will be used to override the loaded values.
- Returns
A model config, which inherited from PretrainedConfig.
- static register(model_type, config, exist_ok=False)[source]
Register a new configuration for this class.
Warning
The API is experimental and may have some slight breaking changes in the next releases.
- Parameters
model_type (str) – The model type like "bert" or "gpt".
config (PretrainedConfig) – The config to register.
exist_ok (bool, optional) – If set to True, no error will be raised even if model_type already exists. Default:
False.