Hugging Face's logo

Transformers.js documentation

generation/configuration_utils

Hugging Face's logo
Join the Hugging Face community

and get access to the augmented documentation experience

to get started

generation/configuration_utils

generation/configuration_utils.GenerationConfig

Class that holds a configuration for a generation task.

Kind: static class of generation/configuration_utils

new GenerationConfig(config)

ParamType
config*

generationConfig.max_length : <code> number </code>

The maximum length the generated tokens can have. Corresponds to the length of the input prompt + max_new_tokens. Its effect is overridden by max_new_tokens, if also set.

Kind: instance property of GenerationConfig
Default: 20

generationConfig.max_new_tokens : <code> number </code>

The maximum numbers of tokens to generate, ignoring the number of tokens in the prompt.

Kind: instance property of GenerationConfig
Default: null

generationConfig.min_length : <code> number </code>

The minimum length of the sequence to be generated. Corresponds to the length of the input prompt + min_new_tokens. Its effect is overridden by min_new_tokens, if also set.

Kind: instance property of GenerationConfig
Default: 0

generationConfig.min_new_tokens : <code> number </code>

The minimum numbers of tokens to generate, ignoring the number of tokens in the prompt.

Kind: instance property of GenerationConfig
Default: null

generationConfig.early_stopping : <code> boolean </code> | <code> ” never ” </code>

Controls the stopping condition for beam-based methods, like beam-search. It accepts the following values:

  • true, where the generation stops as soon as there are num_beams complete candidates;
  • false, where an heuristic is applied and the generation stops when is it very unlikely to find better candidates;
  • "never", where the beam search procedure only stops when there cannot be better candidates (canonical beam search algorithm).

Kind: instance property of GenerationConfig
Default: false

generationConfig.max_time : <code> number </code>

The maximum amount of time you allow the computation to run for in seconds. Generation will still finish the current pass after allocated time has been passed.

Kind: instance property of GenerationConfig
Default: null

generationConfig.do_sample : <code> boolean </code>

Whether or not to use sampling; use greedy decoding otherwise.

Kind: instance property of GenerationConfig
Default: false

generationConfig.num_beams : <code> number </code>

Number of beams for beam search. 1 means no beam search.

Kind: instance property of GenerationConfig
Default: 1

generationConfig.num_beam_groups : <code> number </code>

Number of groups to divide num_beams into in order to ensure diversity among different groups of beams. See this paper for more details.

Kind: instance property of GenerationConfig
Default: 1

generationConfig.penalty_alpha : <code> number </code>

The values balance the model confidence and the degeneration penalty in contrastive search decoding.

Kind: instance property of GenerationConfig
Default: null

generationConfig.use_cache : <code> boolean </code>

Whether or not the model should use the past last key/values attentions (if applicable to the model) to speed up decoding.

Kind: instance property of GenerationConfig
Default: true

generationConfig.temperature : <code> number </code>

The value used to modulate the next token probabilities.

Kind: instance property of GenerationConfig
Default: 1.0

generationConfig.top_k : <code> number </code>

The number of highest probability vocabulary tokens to keep for top-k-filtering.

Kind: instance property of GenerationConfig
Default: 50

generationConfig.top_p : <code> number </code>

If set to float < 1, only the smallest set of most probable tokens with probabilities that add up to top_p or higher are kept for generation.

Kind: instance property of GenerationConfig
Default: 1.0

generationConfig.typical_p : <code> number </code>

Local typicality measures how similar the conditional probability of predicting a target token next is to the expected conditional probability of predicting a random token next, given the partial text already generated. If set to float < 1, the smallest set of the most locally typical tokens with probabilities that add up to typical_p or higher are kept for generation. See this paper for more details.

Kind: instance property of GenerationConfig
Default: 1.0

generationConfig.epsilon_cutoff : <code> number </code>

If set to float strictly between 0 and 1, only tokens with a conditional probability greater than epsilon_cutoff will be sampled. In the paper, suggested values range from 3e-4 to 9e-4, depending on the size of the model. See Truncation Sampling as Language Model Desmoothing for more details.

Kind: instance property of GenerationConfig
Default: 0.0

generationConfig.eta_cutoff : <code> number </code>

Eta sampling is a hybrid of locally typical sampling and epsilon sampling. If set to float strictly between 0 and 1, a token is only considered if it is greater than either eta_cutoff or sqrt(eta_cutoff) * exp(-entropy(softmax(next_token_logits))). The latter term is intuitively the expected next token probability, scaled by sqrt(eta_cutoff). In the paper, suggested values range from 3e-4 to 2e-3, depending on the size of the model. See Truncation Sampling as Language Model Desmoothing for more details.

Kind: instance property of GenerationConfig
Default: 0.0

generationConfig.diversity_penalty : <code> number </code>

This value is subtracted from a beam’s score if it generates a token same as any beam from other group at a particular time. Note that diversity_penalty is only effective if group beam search is enabled.

Kind: instance property of GenerationConfig
Default: 0.0

generationConfig.repetition_penalty : <code> number </code>

The parameter for repetition penalty. 1.0 means no penalty. See this paper for more details.

Kind: instance property of GenerationConfig
Default: 1.0

generationConfig.encoder_repetition_penalty : <code> number </code>

The paramater for encoder_repetition_penalty. An exponential penalty on sequences that are not in the original input. 1.0 means no penalty.

Kind: instance property of GenerationConfig
Default: 1.0

generationConfig.length_penalty : <code> number </code>

Exponential penalty to the length that is used with beam-based generation. It is applied as an exponent to the sequence length, which in turn is used to divide the score of the sequence. Since the score is the log likelihood of the sequence (i.e. negative), length_penalty > 0.0 promotes longer sequences, while length_penalty < 0.0 encourages shorter sequences.

Kind: instance property of GenerationConfig
Default: 1.0

generationConfig.no_repeat_ngram_size : <code> number </code>

If set to int > 0, all ngrams of that size can only occur once.

Kind: instance property of GenerationConfig
Default: 0

generationConfig.bad_words_ids : <code> Array. < Array < number > > </code>

List of token ids that are not allowed to be generated. In order to get the token ids of the words that should not appear in the generated text, use tokenizer(bad_words, { add_prefix_space: true, add_special_tokens: false }).input_ids.

Kind: instance property of GenerationConfig
Default: null

generationConfig.force_words_ids : <code> Array < Array < number > > </code> | <code> Array < Array < Array < number > > > </code>

List of token ids that must be generated. If given a number[][], this is treated as a simple list of words that must be included, the opposite to bad_words_ids. If given number[][][], this triggers a disjunctive constraint, where one can allow different forms of each word.

Kind: instance property of GenerationConfig
Default: null

generationConfig.renormalize_logits : <code> boolean </code>

Whether to renormalize the logits after applying all the logits processors or warpers (including the custom ones). It’s highly recommended to set this flag to true as the search algorithms suppose the score logits are normalized but some logit processors or warpers break the normalization.

Kind: instance property of GenerationConfig
Default: false

generationConfig.constraints : <code> Array. < Object > </code>

Custom constraints that can be added to the generation to ensure that the output will contain the use of certain tokens as defined by Constraint objects, in the most sensible way possible.

Kind: instance property of GenerationConfig
Default: null

generationConfig.forced_bos_token_id : <code> number </code>

The id of the token to force as the first generated token after the decoder_start_token_id. Useful for multilingual models like mBART where the first generated token needs to be the target language token.

Kind: instance property of GenerationConfig
Default: null

generationConfig.forced_eos_token_id : <code> number </code> | <code> Array < number > </code>

The id of the token to force as the last generated token when max_length is reached. Optionally, use a list to set multiple end-of-sequence tokens.

Kind: instance property of GenerationConfig
Default: null

generationConfig.remove_invalid_values : <code> boolean </code>

Whether to remove possible nan and inf outputs of the model to prevent the generation method to crash. Note that using remove_invalid_values can slow down generation.

Kind: instance property of GenerationConfig

generationConfig.exponential_decay_length_penalty : <code> * </code>

This Tuple adds an exponentially increasing length penalty, after a certain amount of tokens have been generated. The tuple shall consist of: (start_index, decay_factor) where start_index indicates where penalty starts and decay_factor represents the factor of exponential decay.

Kind: instance property of GenerationConfig
Default: null

generationConfig.suppress_tokens : <code> Array. < number > </code>

A list of tokens that will be suppressed at generation. The SuppressTokens logit processor will set their log probs to -inf so that they are not sampled.

Kind: instance property of GenerationConfig
Default: null

generationConfig.begin_suppress_tokens : <code> Array. < number > </code>

A list of tokens that will be suppressed at the beginning of the generation. The SuppressBeginTokens logit processor will set their log probs to -inf so that they are not sampled.

Kind: instance property of GenerationConfig
Default: null

generationConfig.forced_decoder_ids : <code> * </code>

A list of pairs of integers which indicates a mapping from generation indices to token indices that will be forced before sampling. For example, [[1, 123]] means the second generated token will always be a token of index 123.

Kind: instance property of GenerationConfig
Default: null

generationConfig.guidance_scale : <code> number </code>

The guidance scale for classifier free guidance (CFG). CFG is enabled by setting guidance_scale > 1. Higher guidance scale encourages the model to generate samples that are more closely linked to the input prompt, usually at the expense of poorer quality.

Kind: instance property of GenerationConfig
Default: null

generationConfig.num_return_sequences : <code> number </code>

The number of independently computed returned sequences for each element in the batch.

Kind: instance property of GenerationConfig
Default: 1

generationConfig.output_attentions : <code> boolean </code>

Whether or not to return the attentions tensors of all attention layers. See attentions under returned tensors for more details.

Kind: instance property of GenerationConfig
Default: false

generationConfig.output_hidden_states : <code> boolean </code>

Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for more details.

Kind: instance property of GenerationConfig
Default: false

generationConfig.output_scores : <code> boolean </code>

Whether or not to return the prediction scores. See scores under returned tensors for more details.

Kind: instance property of GenerationConfig
Default: false

generationConfig.return_dict_in_generate : <code> boolean </code>

Whether or not to return a ModelOutput instead of a plain tuple.

Kind: instance property of GenerationConfig
Default: false

generationConfig.pad_token_id : <code> number </code>

The id of the padding token.

Kind: instance property of GenerationConfig
Default: null

generationConfig.bos_token_id : <code> number </code>

The id of the beginning-of-sequence token.

Kind: instance property of GenerationConfig
Default: null

generationConfig.eos_token_id : <code> number </code> | <code> Array < number > </code>

The id of the end-of-sequence token. Optionally, use a list to set multiple end-of-sequence tokens.

Kind: instance property of GenerationConfig
Default: null

generationConfig.encoder_no_repeat_ngram_size : <code> number </code>

If set to int > 0, all ngrams of that size that occur in the encoder_input_ids cannot occur in the decoder_input_ids.

Kind: instance property of GenerationConfig
Default: 0

generationConfig.decoder_start_token_id : <code> number </code>

If an encoder-decoder model starts decoding with a different token than bos, the id of that token.

Kind: instance property of GenerationConfig
Default: null

generationConfig.generation_kwargs : <code> Object </code>

Additional generation kwargs will be forwarded to the generate function of the model. Kwargs that are not present in generate’s signature will be used in the model forward pass.

Kind: instance property of GenerationConfig
Default: {}

< > Update on GitHub

←Parameters Logits Processors→