transformers.js/pr_1665/en/api/configs.md

configs

Helper module for using model configs. For more information, see the corresponding Python documentation.

Example: Load an AutoConfig.

import { AutoConfig } from '@huggingface/transformers';
const config = await AutoConfig.from_pretrained('bert-base-uncased');
console.log(config);
// PretrainedConfig {
//   "model_type": "bert",
//   "is_encoder_decoder": false,
//   "architectures": [
//       "BertForMaskedLM"
//   ],
//   "vocab_size": 30522
//   "num_attention_heads": 12,
//   "num_hidden_layers": 12,
//   "hidden_size": 768,
//   "max_position_embeddings": 512,
//   ...
// }

Classes

PretrainedConfig

Base class for all configuration classes. For more information, see the corresponding Python documentation.

PretrainedConfig.model_type : string | null

Default: null

PretrainedConfig.is_encoder_decoder : boolean

Default: false

PretrainedConfig.max_position_embeddings : number

PretrainedConfig.'transformers.js_config' : TransformersJSConfig

PretrainedConfig.constructor(configJSON)

Create a new PretrainedConfig from a parsed config.json object.

Parameters

• configJSON (Object) — The JSON of the config.

PretrainedConfig.from_pretrained(pretrained_model_name_or_path, options)

Loads a pretrained config from the given pretrained_model_name_or_path.

Parameters

• pretrained_model_name_or_path (string) — The path to the pretrained config.
• options (PretrainedOptions) — Additional options for loading the config.

Returns: Promise<PretrainedConfig> — A new instance of the PretrainedConfig class.

Throws

• Error — Throws an error if the config.json is not found in the pretrained_model_name_or_path.

AutoConfig

Loads a model config from a pretrained id. Thin alias for PretrainedConfig.from_pretrained.

import { AutoConfig } from '@huggingface/transformers';

const config = await AutoConfig.from_pretrained('Xenova/bert-base-uncased');

AutoConfig.from_pretrained(pretrained_model_name_or_path, options)

Loads a pretrained config from the given pretrained_model_name_or_path.

Parameters

• pretrained_model_name_or_path (string) — The path to the pretrained config.
• options (PretrainedOptions) — Additional options for loading the config.

Returns: Promise<PretrainedConfig> — A new instance of the PretrainedConfig class.

Throws

• Error — Throws an error if the config.json is not found in the pretrained_model_name_or_path.

Type Definitions

TransformersJSConfig

Transformers.js-specific configuration, possibly present in config.json under the key transformers.js_config.

Properties

• device_config (Record<DeviceType, DeviceConfig>) optional — Device-specific configurations.
• free_dimension_overrides (Record<string, number>) optional — Override the free dimensions of the model. See https://onnxruntime.ai/docs/tutorials/web/env-flags-and-session-options.html#freedimensionoverrides for more information.
• device (DeviceType) optional — The default device to use for the model.
• dtype (DataType | Record<string, DataType>) optional — The default data type to use for the model.
• use_external_data_format (ExternalData | Record<string, ExternalData>) optional — Whether to load the model using the external data format (used for models >= 2GB in size).

DeviceConfig

Device-specific configuration options.

Type: Omit<TransformersJSConfig, 'device' | 'device_config'>