Hugging Face's logo

Transformers documentation

Generation

Hugging Face's logo
Join the Hugging Face community

and get access to the augmented documentation experience

to get started

Generation

每个框架都在它们各自的 GenerationMixin 类中实现了文本生成的 generate 方法：

无论您选择哪个框架，都可以使用 GenerationConfig 类实例对 generate 方法进行参数化。有关生成方法的控制参数的完整列表，请参阅此类。

要了解如何检查模型的生成配置、默认值是什么、如何临时更改参数以及如何创建和保存自定义生成配置，请参阅 文本生成策略指南。该指南还解释了如何使用相关功能，如token流。

GenerationConfig

class transformers.GenerationConfig

< >

( **kwargs )

Parameters that control the length of the output

  • max_length (int, optional, defaults to 20) — 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.
  • max_new_tokens (int, optional) — The maximum numbers of tokens to generate, ignoring the number of tokens in the prompt.
  • min_length (int, optional, defaults to 0) — 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.
  • min_new_tokens (int, optional) — The minimum numbers of tokens to generate, ignoring the number of tokens in the prompt.
  • early_stopping (bool or str, optional, defaults to False) — 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).
  • max_time (float, optional) — 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.
  • stop_strings (str or List[str], optional) — A string or a list of strings that should terminate generation if the model outputs them.

Parameters that control the generation strategy used

  • do_sample (bool, optional, defaults to False) — Whether or not to use sampling ; use greedy decoding otherwise.
  • num_beams (int, optional, defaults to 1) — Number of beams for beam search. 1 means no beam search.
  • num_beam_groups (int, optional, defaults to 1) — Number of groups to divide num_beams into in order to ensure diversity among different groups of beams. this paper for more details.
  • penalty_alpha (float, optional) — The values balance the model confidence and the degeneration penalty in contrastive search decoding.
  • dola_layers (str or List[int], optional) — The layers to use for DoLa decoding. If None, DoLa decoding is not used. If a string, it must be one of “low” or “high”, which means using the lower part or higher part of the model layers, respectively. “low” means the first half of the layers up to the first 20 layers, and “high” means the last half of the layers up to the last 20 layers. If a list of integers, it must contain the indices of the layers to use for candidate premature layers in DoLa. The 0-th layer is the word embedding layer of the model. Set to 'low' to improve long-answer reasoning tasks, 'high' to improve short-answer tasks. Check the documentation or the paper for more details.

Parameters that control the cache

  • use_cache (bool, optional, defaults to True) — Whether or not the model should use the past last key/values attentions (if applicable to the model) to speed up decoding.
  • cache_implementation (str, optional, default to None) — Name of the cache class that will be instantiated in generate, for faster decoding. Possible values are:

    • "static": StaticCache
    • "offloaded_static": OffloadedStaticCache
    • "sliding_window": SlidingWindowCache
    • "hybrid": HybridCache
    • "mamba": MambaCache
    • "quantized": QuantizedCache

    We support other cache types, but they must be manually instantiated and passed to generate through the past_key_values argument. See our cache documentation for further information.

  • cache_config (CacheConfig or dict, optional, default to None) — Arguments used in the key-value cache class can be passed in cache_config. Can be passed as a Dict and it will be converted to its repsective CacheConfig internally. Otherwise can be passed as a CacheConfig class matching the indicated cache_implementation.
  • return_legacy_cache (bool, optional, default to True) — Whether to return the legacy or new format of the cache when DynamicCache is used by default.

Parameters for manipulation of the model output logits

  • temperature (float, optional, defaults to 1.0) — The value used to modulate the next token probabilities.
  • top_k (int, optional, defaults to 50) — The number of highest probability vocabulary tokens to keep for top-k-filtering.
  • top_p (float, optional, defaults to 1.0) — 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.
  • min_p (float, optional) — Minimum token probability, which will be scaled by the probability of the most likely token. It must be a value between 0 and 1. Typical values are in the 0.01-0.2 range, comparably selective as setting top_p in the 0.99-0.8 range (use the opposite of normal top_p values).
  • typical_p (float, optional, defaults to 1.0) — 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.
  • epsilon_cutoff (float, optional, defaults to 0.0) — 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.
  • eta_cutoff (float, optional, defaults to 0.0) — 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.
  • diversity_penalty (float, optional, defaults to 0.0) — 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.
  • repetition_penalty (float, optional, defaults to 1.0) — The parameter for repetition penalty. 1.0 means no penalty. See this paper for more details.
  • encoder_repetition_penalty (float, optional, defaults to 1.0) — The paramater for encoder_repetition_penalty. An exponential penalty on sequences that are not in the original input. 1.0 means no penalty.
  • length_penalty (float, optional, defaults to 1.0) — 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.
  • no_repeat_ngram_size (int, optional, defaults to 0) — If set to int > 0, all ngrams of that size can only occur once.
  • bad_words_ids (List[List[int]], optional) — List of list of token ids that are not allowed to be generated. Check NoBadWordsLogitsProcessor for further documentation and examples.
  • force_words_ids (List[List[int]] or List[List[List[int]]], optional) — List of token ids that must be generated. If given a List[List[int]], this is treated as a simple list of words that must be included, the opposite to bad_words_ids. If given List[List[List[int]]], this triggers a disjunctive constraint, where one can allow different forms of each word.
  • renormalize_logits (bool, optional, defaults to False) — Whether to renormalize the logits after applying all the logits processors (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 break the normalization.
  • constraints (List[Constraint], optional) — 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.
  • forced_bos_token_id (int, optional, defaults to model.config.forced_bos_token_id) — 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.
  • forced_eos_token_id (int or List[int], *optional*, defaults to model.config.forced_eos_token_id) -- 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.
  • remove_invalid_values (bool, optional, defaults to model.config.remove_invalid_values) — 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.
  • exponential_decay_length_penalty (tuple(int, float), optional) — 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
  • suppress_tokens (List[int], optional) — A list of tokens that will be suppressed at generation. The SupressTokens logit processor will set their log probs to -inf so that they are not sampled.
  • begin_suppress_tokens (List[int], optional) — A list of tokens that will be suppressed at the beginning of the generation. The SupressBeginTokens logit processor will set their log probs to -inf so that they are not sampled.
  • forced_decoder_ids (List[List[int]], optional) — 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.
  • sequence_bias (Dict[Tuple[int], float], optional)) — Dictionary that maps a sequence of tokens to its bias term. Positive biases increase the odds of the sequence being selected, while negative biases do the opposite. Check SequenceBiasLogitsProcessor for further documentation and examples.
  • token_healing (bool, optional, defaults to False) — Heal tail tokens of prompts by replacing them with their appropriate extensions. This enhances the quality of completions for prompts affected by greedy tokenization bias.
  • guidance_scale (float, optional) — 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.
  • low_memory (bool, optional) — Switch to sequential beam search and sequential topk for contrastive search to reduce peak memory. Used with beam search and contrastive search.
  • watermarking_config (BaseWatermarkingConfig or dict, optional) — Arguments used to watermark the model outputs by adding a small bias to randomly selected set of “green” tokens. See the docs of SynthIDTextWatermarkingConfig and WatermarkingConfig for more details. If passed as Dict, it will be converted to a WatermarkingConfig internally.

Parameters that define the output variables of generate

  • num_return_sequences (int, optional, defaults to 1) — The number of independently computed returned sequences for each element in the batch.
  • output_attentions (bool, optional, defaults to False) — Whether or not to return the attentions tensors of all attention layers. See attentions under returned tensors for more details.
  • output_hidden_states (bool, optional, defaults to False) — Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for more details.
  • output_scores (bool, optional, defaults to False) — Whether or not to return the prediction scores. See scores under returned tensors for more details.
  • output_logits (bool, optional) — Whether or not to return the unprocessed prediction logit scores. See logits under returned tensors for more details.
  • return_dict_in_generate (bool, optional, defaults to False) — Whether or not to return a ModelOutput, as opposed to returning exclusively the generated sequence. This flag must be set to True to return the generation cache (when use_cache is True) or optional outputs (see flags starting with output_)

Special tokens that can be used at generation time

  • pad_token_id (int, optional) — The id of the padding token.
  • bos_token_id (int, optional) — The id of the beginning-of-sequence token.
  • eos_token_id (Union[int, List[int]], optional) — The id of the end-of-sequence token. Optionally, use a list to set multiple end-of-sequence tokens.

Generation parameters exclusive to encoder-decoder models

  • encoder_no_repeat_ngram_size (int, optional, defaults to 0) — If set to int > 0, all ngrams of that size that occur in the encoder_input_ids cannot occur in the decoder_input_ids.
  • decoder_start_token_id (int or List[int], optional) — If an encoder-decoder model starts decoding with a different token than bos, the id of that token or a list of length batch_size. Indicating a list enables different start ids for each element in the batch (e.g. multilingual models with different target languages in one batch)

Generation parameters exclusive to assistant generation

  • is_assistant (bool, optional, defaults to False) — Whether the model is an assistant (draft) model.
  • num_assistant_tokens (int, optional, defaults to 20) — Defines the number of speculative tokens that shall be generated by the assistant model before being checked by the target model at each iteration. Higher values for num_assistant_tokens make the generation more speculative : If the assistant model is performant larger speed-ups can be reached, if the assistant model requires lots of corrections, lower speed-ups are reached.
  • num_assistant_tokens_schedule (str, optional, defaults to "constant") — Defines the schedule at which max assistant tokens shall be changed during inference.
    • "heuristic": When all speculative tokens are correct, increase num_assistant_tokens by 2 else reduce by 1. num_assistant_tokens value is persistent over multiple generation calls with the same assistant model.
    • "heuristic_transient": Same as "heuristic" but num_assistant_tokens is reset to its initial value after each generation call.
    • "constant": num_assistant_tokens stays unchanged during generation
  • assistant_confidence_threshold (float, optional, defaults to 0.4) — The confidence threshold for the assistant model. If the assistant model’s confidence in its prediction for the current token is lower than this threshold, the assistant model stops the current token generation iteration, even if the number of speculative tokens (defined by num_assistant_tokens) is not yet reached. The assistant’s confidence threshold is adjusted throughout the speculative iterations to reduce the number of unnecessary draft and target forward passes, biased towards avoiding false negatives. assistant_confidence_threshold value is persistent over multiple generation calls with the same assistant model. It is an unsupervised version of the dynamic speculation lookahead from Dynamic Speculation Lookahead Accelerates Speculative Decoding of Large Language Models https://arxiv.org/abs/2405.04304.
  • prompt_lookup_num_tokens (int, optional) — The number of tokens to be output as candidate tokens.
  • max_matching_ngram_size (int, optional) — The maximum ngram size to be considered for matching in the prompt. Default to 2 if not provided.
  • assistant_early_exit(int, optional) — If set to a positive integer, early exit of the model will be used as an assistant. Can only be used with models that support early exit (i.e. models where logits from intermediate layers can be interpreted by the LM head).
  • assistant_lookbehind(int, optional, defaults to 10) — If set to a positive integer, the re-encodeing process will additionally consider the last assistant_lookbehind assistant tokens to correctly align tokens. Can only be used with different tokenizers in speculative decoding. See this blog for more details.
  • target_lookbehind(int, optional, defaults to 10) — If set to a positive integer, the re-encodeing process will additionally consider the last target_lookbehind target tokens to correctly align tokens. Can only be used with different tokenizers in speculative decoding. See this blog for more details.

Parameters related to performances and compilation

  • compile_config (CompileConfig, optional) — If using a static cache, this controls how generate will compile the forward pass for performance gains.

Wild card

  • generation_kwargs — 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.

Class that holds a configuration for a generation task. A generate call supports the following generation methods for text-decoder, text-to-text, speech-to-text, and vision-to-text models:

  • greedy decoding if num_beams=1 and do_sample=False
  • contrastive search if penalty_alpha>0. and top_k>1
  • multinomial sampling if num_beams=1 and do_sample=True
  • beam-search decoding if num_beams>1 and do_sample=False
  • beam-search multinomial sampling if num_beams>1 and do_sample=True
  • diverse beam-search decoding if num_beams>1 and num_beam_groups>1
  • constrained beam-search decoding if constraints!=None or force_words_ids!=None
  • assisted decoding if assistant_model or prompt_lookup_num_tokens is passed to .generate()
  • dola decoding if dola_layers is passed to .generate()

To learn more about decoding strategies refer to the text generation strategies guide.

A large number of these flags control the logits or the stopping criteria of the generation. Make sure you check the generate-related classes for a full description of the possible manipulations, as well as examples of their usage.

from_pretrained

< >

( pretrained_model_name: typing.Union[str, os.PathLike] config_file_name: typing.Union[str, os.PathLike, NoneType] = None cache_dir: typing.Union[str, os.PathLike, NoneType] = None force_download: bool = False local_files_only: bool = False token: typing.Union[str, bool, NoneType] = None revision: str = 'main' **kwargs ) → GenerationConfig

Parameters

  • pretrained_model_name (str or os.PathLike) — This can be either:

    • a string, the model id of a pretrained model configuration hosted inside a model repo on huggingface.co.
    • a path to a directory containing a configuration file saved using the save_pretrained() method, e.g., ./my_model_directory/.
  • config_file_name (str or os.PathLike, optional, defaults to "generation_config.json") — Name of the generation configuration JSON file to be loaded from pretrained_model_name.
  • cache_dir (str or os.PathLike, optional) — Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used.
  • force_download (bool, optional, defaults to False) — Whether or not to force to (re-)download the configuration files and override the cached versions if they exist.
  • resume_download — Deprecated and ignored. All downloads are now resumed by default when possible. Will be removed in v5 of Transformers.
  • proxies (Dict[str, str], optional) — A dictionary of proxy servers to use by protocol or endpoint, e.g., {'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}. The proxies are used on each request.
  • token (str or bool, optional) — The token to use as HTTP bearer authorization for remote files. If True, or not specified, will use the token generated when running huggingface-cli login (stored in ~/.huggingface).
  • revision (str, optional, defaults to "main") — The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so revision can be any identifier allowed by git.

    To test a pull request you made on the Hub, you can pass revision="refs/pr/<pr_number>".
  • return_unused_kwargs (bool, optional, defaults to False) — If False, then this function returns just the final configuration object.

    If True, then this functions returns a Tuple(config, unused_kwargs) where unused_kwargs is a dictionary consisting of the key/value pairs whose keys are not configuration attributes: i.e., the part of kwargs which has not been used to update config and is otherwise ignored.

  • subfolder (str, optional, defaults to "") — In case the relevant files are located inside a subfolder of the model repo on huggingface.co, you can specify the folder name here.
  • kwargs (Dict[str, Any], optional) — The values in kwargs of any keys which are configuration attributes will be used to override the loaded values. Behavior concerning key/value pairs whose keys are not configuration attributes is controlled by the return_unused_kwargs keyword parameter.

Returns

GenerationConfig

The configuration object instantiated from this pretrained model.

Instantiate a GenerationConfig from a generation configuration file.

Examples:

>>> from transformers import GenerationConfig

>>> # Download configuration from huggingface.co and cache.
>>> generation_config = GenerationConfig.from_pretrained("openai-community/gpt2")

>>> # E.g. config was saved using *save_pretrained('./test/saved_model/')*
>>> generation_config.save_pretrained("./test/saved_model/")
>>> generation_config = GenerationConfig.from_pretrained("./test/saved_model/")

>>> # You can also specify configuration names to your generation configuration file
>>> generation_config.save_pretrained("./test/saved_model/", config_file_name="my_configuration.json")
>>> generation_config = GenerationConfig.from_pretrained("./test/saved_model/", "my_configuration.json")

>>> # If you'd like to try a minor variation to an existing configuration, you can also pass generation
>>> # arguments to `.from_pretrained()`. Be mindful that typos and unused arguments will be ignored
>>> generation_config, unused_kwargs = GenerationConfig.from_pretrained(
...     "openai-community/gpt2", top_k=1, foo=False, do_sample=True, return_unused_kwargs=True
... )
>>> generation_config.top_k
1

>>> unused_kwargs
{'foo': False}

from_model_config

< >

( model_config: PretrainedConfig ) → GenerationConfig

Parameters

  • model_config (PretrainedConfig) — The model config that will be used to instantiate the generation config.

Returns

GenerationConfig

The configuration object instantiated from those parameters.

Instantiates a GenerationConfig from a PretrainedConfig. This function is useful to convert legacy PretrainedConfig objects, which may contain generation parameters, into a stand-alone GenerationConfig.

save_pretrained

< >

( save_directory: typing.Union[str, os.PathLike] config_file_name: typing.Union[str, os.PathLike, NoneType] = None push_to_hub: bool = False **kwargs )

Parameters

  • save_directory (str or os.PathLike) — Directory where the configuration JSON file will be saved (will be created if it does not exist).
  • config_file_name (str or os.PathLike, optional, defaults to "generation_config.json") — Name of the generation configuration JSON file to be saved in save_directory.
  • push_to_hub (bool, optional, defaults to False) — Whether or not to push your model to the Hugging Face model hub after saving it. You can specify the repository you want to push to with repo_id (will default to the name of save_directory in your namespace).
  • kwargs (Dict[str, Any], optional) — Additional key word arguments passed along to the push_to_hub() method.

Save a generation configuration object to the directory save_directory, so that it can be re-loaded using the from_pretrained() class method.

GenerationMixin

class transformers.GenerationMixin

< >

( )

A class containing all functions for auto-regressive text generation, to be used as a mixin in PreTrainedModel.

The class exposes generate(), which can be used for:

  • greedy decoding if num_beams=1 and do_sample=False
  • contrastive search if penalty_alpha>0 and top_k>1
  • multinomial sampling if num_beams=1 and do_sample=True
  • beam-search decoding if num_beams>1 and do_sample=False
  • beam-search multinomial sampling if num_beams>1 and do_sample=True
  • diverse beam-search decoding if num_beams>1 and num_beam_groups>1
  • constrained beam-search decoding if constraints!=None or force_words_ids!=None
  • assisted decoding if assistant_model or prompt_lookup_num_tokens is passed to .generate()

To learn more about decoding strategies refer to the text generation strategies guide.

generate

< >

( inputs: typing.Optional[torch.Tensor] = None generation_config: typing.Optional[transformers.generation.configuration_utils.GenerationConfig] = None logits_processor: typing.Optional[transformers.generation.logits_process.LogitsProcessorList] = None stopping_criteria: typing.Optional[transformers.generation.stopping_criteria.StoppingCriteriaList] = None prefix_allowed_tokens_fn: typing.Optional[typing.Callable[[int, torch.Tensor], typing.List[int]]] = None synced_gpus: typing.Optional[bool] = None assistant_model: typing.Optional[ForwardRef('PreTrainedModel')] = None streamer: typing.Optional[ForwardRef('BaseStreamer')] = None negative_prompt_ids: typing.Optional[torch.Tensor] = None negative_prompt_attention_mask: typing.Optional[torch.Tensor] = None **kwargs ) → ModelOutput or torch.LongTensor

Parameters

  • inputs (torch.Tensor of varying shape depending on the modality, optional) — The sequence used as a prompt for the generation or as model inputs to the encoder. If None the method initializes it with bos_token_id and a batch size of 1. For decoder-only models inputs should be in the format of input_ids. For encoder-decoder models inputs can represent any of input_ids, input_values, input_features, or pixel_values.
  • generation_config (GenerationConfig, optional) — The generation configuration to be used as base parametrization for the generation call. **kwargs passed to generate matching the attributes of generation_config will override them. If generation_config is not provided, the default will be used, which has the following loading priority: 1) from the generation_config.json model file, if it exists; 2) from the model configuration. Please note that unspecified parameters will inherit GenerationConfig’s default values, whose documentation should be checked to parameterize generation.
  • logits_processor (LogitsProcessorList, optional) — Custom logits processors that complement the default logits processors built from arguments and generation config. If a logit processor is passed that is already created with the arguments or a generation config an error is thrown. This feature is intended for advanced users.
  • stopping_criteria (StoppingCriteriaList, optional) — Custom stopping criteria that complements the default stopping criteria built from arguments and a generation config. If a stopping criteria is passed that is already created with the arguments or a generation config an error is thrown. If your stopping criteria depends on the scores input, make sure you pass return_dict_in_generate=True, output_scores=True to generate. This feature is intended for advanced users.
  • prefix_allowed_tokens_fn (Callable[[int, torch.Tensor], List[int]], optional) — If provided, this function constraints the beam search to allowed tokens only at each step. If not provided no constraint is applied. This function takes 2 arguments: the batch ID batch_id and input_ids. It has to return a list with the allowed tokens for the next generation step conditioned on the batch ID batch_id and the previously generated tokens inputs_ids. This argument is useful for constrained generation conditioned on the prefix, as described in Autoregressive Entity Retrieval.
  • synced_gpus (bool, optional) — Whether to continue running the while loop until max_length. Unless overridden, this flag will be set to True if using FullyShardedDataParallel or DeepSpeed ZeRO Stage 3 with multiple GPUs to avoid deadlocking if one GPU finishes generating before other GPUs. Otherwise, defaults to False.
  • assistant_model (PreTrainedModel, optional) — An assistant model that can be used to accelerate generation. The assistant model must have the exact same tokenizer. The acceleration is achieved when forecasting candidate tokens with the assistant model is much faster than running generation with the model you’re calling generate from. As such, the assistant model should be much smaller.
  • streamer (BaseStreamer, optional) — Streamer object that will be used to stream the generated sequences. Generated tokens are passed through streamer.put(token_ids) and the streamer is responsible for any further processing.
  • negative_prompt_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) — The negative prompt needed for some processors such as CFG. The batch size must match the input batch size. This is an experimental feature, subject to breaking API changes in future versions.
  • negative_prompt_attention_mask (torch.LongTensor of shape (batch_size, sequence_length), optional) — Attention_mask for negative_prompt_ids.
  • kwargs (Dict[str, Any], optional) — Ad hoc parametrization of generation_config and/or additional model-specific kwargs that will be forwarded to the forward function of the model. If the model is an encoder-decoder model, encoder specific kwargs should not be prefixed and decoder specific kwargs should be prefixed with decoder_.

Returns

ModelOutput or torch.LongTensor

A ModelOutput (if return_dict_in_generate=True or when config.return_dict_in_generate=True) or a torch.LongTensor.

If the model is not an encoder-decoder model (model.config.is_encoder_decoder=False), the possible ModelOutput types are:

If the model is an encoder-decoder model (model.config.is_encoder_decoder=True), the possible ModelOutput types are:

Generates sequences of token ids for models with a language modeling head.

Most generation-controlling parameters are set in generation_config which, if not passed, will be set to the model’s default generation configuration. You can override any generation_config by passing the corresponding parameters to generate(), e.g. .generate(inputs, num_beams=4, do_sample=True).

For an overview of generation strategies and code examples, check out the following guide.

compute_transition_scores

< >

( sequences: Tensor scores: typing.Tuple[torch.Tensor] beam_indices: typing.Optional[torch.Tensor] = None normalize_logits: bool = False ) → torch.Tensor

Parameters

  • sequences (torch.LongTensor) — The generated sequences. The second dimension (sequence_length) is either equal to max_length or shorter if all batches finished early due to the eos_token_id.
  • scores (tuple(torch.FloatTensor)) — Transition scores for each vocabulary token at each generation step. Beam transition scores consisting of log probabilities of tokens conditioned on log softmax of previously generated tokens in this beam. Tuple of torch.FloatTensor with up to max_new_tokens elements (one element for each generated token), with each tensor of shape (batch_size*num_beams, config.vocab_size).
  • beam_indices (torch.LongTensor, optional) — Beam indices of generated token id at each generation step. torch.LongTensor of shape (batch_size*num_return_sequences, sequence_length). Only required if a num_beams>1 at generate-time.
  • normalize_logits (bool, optional, defaults to False) — Whether to normalize the logits (which, for legacy reasons, may be unnormalized).

Returns

torch.Tensor

A torch.Tensor of shape (batch_size*num_return_sequences, sequence_length) containing the transition scores (logits)

Computes the transition scores of sequences given the generation scores (and beam indices, if beam search was used). This is a convenient method to quicky obtain the scores of the selected tokens at generation time.

Examples:

>>> from transformers import GPT2Tokenizer, AutoModelForCausalLM
>>> import numpy as np

>>> tokenizer = GPT2Tokenizer.from_pretrained("gpt2")
>>> model = AutoModelForCausalLM.from_pretrained("openai-community/gpt2")
>>> tokenizer.pad_token_id = tokenizer.eos_token_id
>>> inputs = tokenizer(["Today is"], return_tensors="pt")

>>> # Example 1: Print the scores for each token generated with Greedy Search
>>> outputs = model.generate(**inputs, max_new_tokens=5, return_dict_in_generate=True, output_scores=True)
>>> transition_scores = model.compute_transition_scores(
...     outputs.sequences, outputs.scores, normalize_logits=True
... )
>>> # input_length is the length of the input prompt for decoder-only models, like the GPT family, and 1 for
>>> # encoder-decoder models, like BART or T5.
>>> input_length = 1 if model.config.is_encoder_decoder else inputs.input_ids.shape[1]
>>> generated_tokens = outputs.sequences[:, input_length:]
>>> for tok, score in zip(generated_tokens[0], transition_scores[0]):
...     # | token | token string | log probability | probability
...     print(f"| {tok:5d} | {tokenizer.decode(tok):8s} | {score.numpy():.3f} | {np.exp(score.numpy()):.2%}")
|   262 |  the     | -1.414 | 24.33%
|  1110 |  day     | -2.609 | 7.36%
|   618 |  when    | -2.010 | 13.40%
|   356 |  we      | -1.859 | 15.58%
|   460 |  can     | -2.508 | 8.14%

>>> # Example 2: Reconstruct the sequence scores from Beam Search
>>> outputs = model.generate(
...     **inputs,
...     max_new_tokens=5,
...     num_beams=4,
...     num_return_sequences=4,
...     return_dict_in_generate=True,
...     output_scores=True,
... )
>>> transition_scores = model.compute_transition_scores(
...     outputs.sequences, outputs.scores, outputs.beam_indices, normalize_logits=False
... )
>>> # If you sum the generated tokens' scores and apply the length penalty, you'll get the sequence scores.
>>> # Tip 1: recomputing the scores is only guaranteed to match with `normalize_logits=False`. Depending on the
>>> # use case, you might want to recompute it with `normalize_logits=True`.
>>> # Tip 2: the output length does NOT include the input length
>>> output_length = np.sum(transition_scores.numpy() < 0, axis=1)
>>> length_penalty = model.generation_config.length_penalty
>>> reconstructed_scores = transition_scores.sum(axis=1) / (output_length**length_penalty)
>>> print(np.allclose(outputs.sequences_scores, reconstructed_scores))
True

TFGenerationMixin

class transformers.TFGenerationMixin

< >

( )

A class containing all of the functions supporting generation, to be used as a mixin in TFPreTrainedModel.

The class exposes generate(), which can be used for:

  • greedy decoding by calling greedy_search() if num_beams=1 and do_sample=False
  • contrastive search by calling contrastive_search() if penalty_alpha>0 and top_k>1
  • multinomial sampling by calling sample() if num_beams=1 and do_sample=True
  • beam-search decoding by calling beam_search() if num_beams>1

You do not need to call any of the above methods directly. Pass custom parameter values to ‘generate’ instead. To learn more about decoding strategies refer to the text generation strategies guide.

generate

< >

( inputs: typing.Optional[tensorflow.python.framework.tensor.Tensor] = None generation_config: typing.Optional[transformers.generation.configuration_utils.GenerationConfig] = None logits_processor: typing.Optional[transformers.generation.tf_logits_process.TFLogitsProcessorList] = None seed = None **kwargs ) → ModelOutput or tf.Tensor

Parameters

  • inputs (tf.Tensor of varying shape depending on the modality, optional) — The sequence used as a prompt for the generation or as model inputs to the encoder. If None the method initializes it with bos_token_id and a batch size of 1. For decoder-only models inputs should of in the format of input_ids. For encoder-decoder models inputs can represent any of input_ids, input_values, input_features, or pixel_values.
  • generation_config (~generation.GenerationConfig, optional) — The generation configuration to be used as base parametrization for the generation call. **kwargs passed to generate matching the attributes of generation_config will override them. If generation_config is not provided, the default will be used, which had the following loading priority: 1) from the generation_config.json model file, if it exists; 2) from the model configuration. Please note that unspecified parameters will inherit GenerationConfig’s default values, whose documentation should be checked to parameterize generation.
  • logits_processor (LogitsProcessorList, optional) — Custom logits processors that complement the default logits processors built from arguments and generation config. If a logit processor is passed that is already created with the arguments or a generation config an error is thrown. This feature is intended for advanced users.
  • seed (List[int], optional) — Random seed to control sampling, containing two integers, used when do_sample is True. See the seed argument from stateless functions in tf.random.
  • kwargs (Dict[str, Any], optional) — Ad hoc parametrization of generate_config and/or additional model-specific kwargs that will be forwarded to the forward function of the model. If the model is an encoder-decoder model, encoder specific kwargs should not be prefixed and decoder specific kwargs should be prefixed with decoder_.

Returns

ModelOutput or tf.Tensor

A ModelOutput (if return_dict_in_generate=True or when config.return_dict_in_generate=True) or a tf.Tensor.

If the model is not an encoder-decoder model (model.config.is_encoder_decoder=False), the possible ModelOutput types are:

If the model is an encoder-decoder model (model.config.is_encoder_decoder=True), the possible ModelOutput types are:

Generates sequences of token ids for models with a language modeling head.

Most generation-controlling parameters are set in generation_config which, if not passed, will be set to the model’s default generation configuration. You can override any generation_config by passing the corresponding parameters to generate, e.g. .generate(inputs, num_beams=4, do_sample=True).

For an overview of generation strategies and code examples, check out the following guide.

compute_transition_scores

< >

( sequences: Tensor scores: typing.Tuple[tensorflow.python.framework.tensor.Tensor] beam_indices: typing.Optional[tensorflow.python.framework.tensor.Tensor] = None normalize_logits: bool = False ) → tf.Tensor

Parameters

  • sequences (tf.Tensor) — The generated sequences. The second dimension (sequence_length) is either equal to max_length or shorter if all batches finished early due to the eos_token_id.
  • scores (tuple(tf.Tensor)) — Transition scores for each vocabulary token at each generation step. Beam transition scores consisting of log probabilities of tokens conditioned on log softmax of previously generated tokens Tuple of tf.Tensor with up to max_new_tokens elements (one element for each generated token), with each tensor of shape (batch_size*num_beams, config.vocab_size).
  • beam_indices (tf.Tensor, optional) — Beam indices of generated token id at each generation step. tf.Tensor of shape (batch_size*num_return_sequences, sequence_length). Only required if a num_beams>1 at generate-time.
  • normalize_logits (bool, optional, defaults to False) — Whether to normalize the logits (which, for legacy reasons, may be unnormalized).

Returns

tf.Tensor

A tf.Tensor of shape (batch_size*num_return_sequences, sequence_length) containing the transition scores (logits)

Computes the transition scores of sequences given the generation scores (and beam indices, if beam search was used). This is a convenient method to quicky obtain the scores of the selected tokens at generation time.

Examples:

>>> from transformers import GPT2Tokenizer, TFAutoModelForCausalLM
>>> import numpy as np

>>> tokenizer = GPT2Tokenizer.from_pretrained("openai-community/gpt2")
>>> model = TFAutoModelForCausalLM.from_pretrained("openai-community/gpt2")
>>> tokenizer.pad_token_id = tokenizer.eos_token_id
>>> inputs = tokenizer(["Today is"], return_tensors="tf")

>>> # Example 1: Print the scores for each token generated with Greedy Search
>>> outputs = model.generate(**inputs, max_new_tokens=5, return_dict_in_generate=True, output_scores=True)
>>> transition_scores = model.compute_transition_scores(
...     outputs.sequences, outputs.scores, normalize_logits=True
... )
>>> # input_length is the length of the input prompt for decoder-only models, like the GPT family, and 1 for
>>> # encoder-decoder models, like BART or T5.
>>> input_length = 1 if model.config.is_encoder_decoder else inputs.input_ids.shape[1]
>>> generated_tokens = outputs.sequences[:, input_length:]
>>> for tok, score in zip(generated_tokens[0], transition_scores[0]):
...     # | token | token string | logits | probability
...     print(f"| {tok:5d} | {tokenizer.decode(tok):8s} | {score.numpy():.3f} | {np.exp(score.numpy()):.2%}")
|   262 |  the     | -1.414 | 24.33%
|  1110 |  day     | -2.609 | 7.36%
|   618 |  when    | -2.010 | 13.40%
|   356 |  we      | -1.859 | 15.58%
|   460 |  can     | -2.508 | 8.14%

>>> # Example 2: Reconstruct the sequence scores from Beam Search
>>> outputs = model.generate(
...     **inputs,
...     max_new_tokens=5,
...     num_beams=4,
...     num_return_sequences=4,
...     return_dict_in_generate=True,
...     output_scores=True,
... )
>>> transition_scores = model.compute_transition_scores(
...     outputs.sequences, outputs.scores, outputs.beam_indices, normalize_logits=False
... )
>>> # If you sum the generated tokens' scores and apply the length penalty, you'll get the sequence scores.
>>> # Tip: recomputing the scores is only guaranteed to match with `normalize_logits=False`. Depending on the
>>> # use case, you might want to recompute it with `normalize_logits=True`.
>>> output_length = np.sum(transition_scores.numpy() < 0, axis=1)
>>> length_penalty = model.generation_config.length_penalty
>>> reconstructed_scores = np.sum(transition_scores, axis=1) / (output_length**length_penalty)
>>> print(np.allclose(outputs.sequences_scores, reconstructed_scores))
True

FlaxGenerationMixin

class transformers.FlaxGenerationMixin

< >

( )

A class containing all functions for auto-regressive text generation, to be used as a mixin in FlaxPreTrainedModel.

The class exposes generate(), which can be used for:

  • greedy decoding by calling _greedy_search() if num_beams=1 and do_sample=False
  • multinomial sampling by calling _sample() if num_beams=1 and do_sample=True
  • beam-search decoding by calling _beam_search() if num_beams>1 and do_sample=False

You do not need to call any of the above methods directly. Pass custom parameter values to ‘generate’ instead. To learn more about decoding strategies refer to the text generation strategies guide.

generate

< >

( input_ids: Array generation_config: typing.Optional[transformers.generation.configuration_utils.GenerationConfig] = None prng_key: typing.Optional[jax.Array] = None trace: bool = True params: typing.Optional[typing.Dict[str, jax.Array]] = None logits_processor: typing.Optional[transformers.generation.flax_logits_process.FlaxLogitsProcessorList] = None **kwargs )

Parameters

  • input_ids (jnp.ndarray of shape (batch_size, sequence_length)) — The sequence used as a prompt for the generation.
  • generation_config (~generation.GenerationConfig, optional) — The generation configuration to be used as base parametrization for the generation call. **kwargs passed to generate matching the attributes of generation_config will override them. If generation_config is not provided, the default will be used, which had the following loading priority: 1) from the generation_config.json model file, if it exists; 2) from the model configuration. Please note that unspecified parameters will inherit GenerationConfig’s default values, whose documentation should be checked to parameterize generation.
  • trace (bool, optional, defaults to True) — Whether to trace generation. Setting trace=False should only be used for debugging and will lead to a considerably slower runtime.
  • params (Dict[str, jnp.ndarray], optional) — Optionally the model parameters can be passed. Can be useful for parallelized generation.
  • logits_processor (FlaxLogitsProcessorList , optional) — Custom logits processors that complement the default logits processors built from arguments and generation config. If a logit processor is passed that is already created with the arguments or a generation config an error is thrown. This feature is intended for advanced users.
  • kwargs (Dict[str, Any], optional) — Ad hoc parametrization of generate_config and/or additional model-specific kwargs that will be forwarded to the forward function of the model. If the model is an encoder-decoder model, encoder specific kwargs should not be prefixed and decoder specific kwargs should be prefixed with decoder_.

Generates sequences of token ids for models with a language modeling head.

< > Update on GitHub

←模型 ONNX→