transformers.js/pr_1649/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,
//   ...
// }

• configs
  • static
    • .PretrainedConfig
      • new PretrainedConfig(configJSON)
      • instance
        • .model_type : string | null
        • .is_encoder_decoder : boolean
        • .max_position_embeddings : number
      • static
        • .from_pretrained(pretrained_model_name_or_path, options) ⇒ Promise.<PretrainedConfig>
    • .AutoConfig
      • new AutoConfig()
      • .from_pretrained() : Object.from_pretrained
    • .getCacheShapes(config) ⇒ Record.<string, Array>
      • ~cache_values : Record.<string, Array>
      • ~cache_values : Record.<string, Array>
      • ~cache_values : Record.<string, Array>
      • ~cache_values : Record.<string, Array>
  • inner
    • ~loadConfig(pretrained_model_name_or_path, options) ⇒ Promise.<Object>
    • ~getNormalizedConfig(config) ⇒ Object
    • ~getKeyValueShapes() : Object
      • ~decoderFeeds : Record.<string, Array>
    • ~PretrainedOptions : PretrainedOptions
    • ~ProgressCallback : ProgressCallback
    • ~ProgressInfo : ProgressInfo

---

configs.PretrainedConfig

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

Kind: static class of configs

• .PretrainedConfig
  • new PretrainedConfig(configJSON)
  • instance
    • .model_type : string | null
    • .is_encoder_decoder : boolean
    • .max_position_embeddings : number
  • static
    • .from_pretrained(pretrained_model_name_or_path, options) ⇒ Promise.<PretrainedConfig>

---

new PretrainedConfig(configJSON)

Create a new PreTrainedTokenizer instance.

  ParamTypeDescription




configJSONObjectThe JSON of the config.

  

---

pretrainedConfig.model_type : string | null

Kind: instance property of PretrainedConfig

---

pretrainedConfig.is_encoder_decoder : boolean

Kind: instance property of PretrainedConfig

---

pretrainedConfig.max_position_embeddings : number

Kind: instance property of PretrainedConfig

---

PretrainedConfig.from_pretrained(pretrained_model_name_or_path, options) ⇒ Promise.<PretrainedConfig>

Loads a pre-trained config from the given pretrained_model_name_or_path.

Kind: static method of PretrainedConfig
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.

  ParamTypeDescription
  

  pretrained_model_name_or_pathstringThe path to the pre-trained config.

  optionsPretrainedOptionsAdditional options for loading the config.

---

configs.AutoConfig

Helper class which is used to instantiate pretrained configs with the from_pretrained function.

Kind: static class of configs

• .AutoConfig
  • new AutoConfig()
  • .from_pretrained() : Object.from_pretrained

---

new AutoConfig()

Example

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

---

AutoConfig.from_pretrained() : Object.from_pretrained

Kind: static method of AutoConfig

---

configs.getCacheShapes(config) ⇒ Record.<string, Array>

Kind: static method of configs

  ParamType




configPretrainedConfig
  

• .getCacheShapes(config) ⇒ Record.<string, Array>
  • ~cache_values : Record.<string, Array>
  • ~cache_values : Record.<string, Array>
  • ~cache_values : Record.<string, Array>
  • ~cache_values : Record.<string, Array>

---

getCacheShapes~cache_values : Record.<string, Array>

Kind: inner constant of getCacheShapes

---

getCacheShapes~cache_values : Record.<string, Array>

Kind: inner constant of getCacheShapes

---

getCacheShapes~cache_values : Record.<string, Array>

Kind: inner constant of getCacheShapes

---

getCacheShapes~cache_values : Record.<string, Array>

Kind: inner constant of getCacheShapes

---

configs~loadConfig(pretrained_model_name_or_path, options) ⇒ Promise.<Object>

Loads a config from the specified path.

Kind: inner method of configs
Returns: Promise.<Object> - A promise that resolves with information about the loaded config.

  ParamTypeDescription




pretrained_model_name_or_pathstringThe path to the config directory.


optionsPretrainedOptionsAdditional options for loading the config.

  

---

configs~getNormalizedConfig(config) ⇒ Object

Kind: inner method of configs
Returns: Object - The normalized configuration.

  ParamType




configPretrainedConfig
  

---

configs~getKeyValueShapes() : Object

Kind: inner method of configs

---

getKeyValueShapes~decoderFeeds : Record.<string, Array>

Kind: inner constant of getKeyValueShapes

---

configs~PretrainedOptions : PretrainedOptions

Kind: inner typedef of configs

---

configs~ProgressCallback : ProgressCallback

Kind: inner typedef of configs

---

configs~ProgressInfo : ProgressInfo

Kind: inner typedef of configs

---