Transformers documentation

RAG

Join the Hugging Face community

to get started

# RAG

## Overview

Retrieval-augmented generation (“RAG”) models combine the powers of pretrained dense retrieval (DPR) and sequence-to-sequence models. RAG models retrieve documents, pass them to a seq2seq model, then marginalize to generate outputs. The retriever and seq2seq modules are initialized from pretrained models, and fine-tuned jointly, allowing both retrieval and generation to adapt to downstream tasks.

It is based on the paper Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks by Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela.

The abstract from the paper is the following:

Large pre-trained language models have been shown to store factual knowledge in their parameters, and achieve state-of-the-art results when fine-tuned on downstream NLP tasks. However, their ability to access and precisely manipulate knowledge is still limited, and hence on knowledge-intensive tasks, their performance lags behind task-specific architectures. Additionally, providing provenance for their decisions and updating their world knowledge remain open research problems. Pre-trained models with a differentiable access mechanism to explicit nonparametric memory can overcome this issue, but have so far been only investigated for extractive downstream tasks. We explore a general-purpose fine-tuning recipe for retrieval-augmented generation (RAG) — models which combine pre-trained parametric and non-parametric memory for language generation. We introduce RAG models where the parametric memory is a pre-trained seq2seq model and the non-parametric memory is a dense vector index of Wikipedia, accessed with a pre-trained neural retriever. We compare two RAG formulations, one which conditions on the same retrieved passages across the whole generated sequence, the other can use different passages per token. We fine-tune and evaluate our models on a wide range of knowledge-intensive NLP tasks and set the state-of-the-art on three open domain QA tasks, outperforming parametric seq2seq models and task-specific retrieve-and-extract architectures. For language generation tasks, we find that RAG models generate more specific, diverse and factual language than a state-of-the-art parametric-only seq2seq baseline.

This model was contributed by ola13.

## RagConfig

### class transformers.RagConfig

< >

( vocab_size = None is_encoder_decoder = True prefix = None bos_token_id = None pad_token_id = None eos_token_id = None decoder_start_token_id = None title_sep = ' / ' doc_sep = ' // ' n_docs = 5 max_combined_length = 300 retrieval_vector_size = 768 retrieval_batch_size = 8 dataset = 'wiki_dpr' dataset_split = 'train' index_name = 'compressed' index_path = None passages_path = None use_dummy_dataset = False reduce_loss = False label_smoothing = 0.0 do_deduplication = True exclude_bos_score = False do_marginalize = False output_retrieved = False use_cache = True forced_eos_token_id = None **kwargs )

Parameters

• title_sep (str, optional, defaults to " / ") — Separator inserted between the title and the text of the retrieved document when calling RagRetriever.
• doc_sep (str, optional, defaults to " // ") — Separator inserted between the the text of the retrieved document and the original input when calling RagRetriever.
• n_docs (int, optional, defaults to 5) — Number of documents to retrieve.
• max_combined_length (int, optional, defaults to 300) — Max length of contextualized input returned by __call__()
• retrieval_vector_size (int, optional, defaults to 768) — Dimensionality of the document embeddings indexed by RagRetriever.
• retrieval_batch_size (int, optional, defaults to 8) — Retrieval batch size, defined as the number of queries issues concurrently to the faiss index encapsulated RagRetriever.
• dataset (str, optional, defaults to "wiki_dpr") — A dataset identifier of the indexed dataset in HuggingFace Datasets (list all available datasets and ids using datasets.list_datasets()).
• dataset_split (str, optional, defaults to "train") — Which split of the dataset to load.
• index_name (str, optional, defaults to "compressed") — The index name of the index associated with the dataset. One can choose between "legacy", "exact" and "compressed".
• index_path (str, optional) — The path to the serialized faiss index on disk. passages_path — (str, optional): A path to text passages compatible with the faiss index. Required if using LegacyIndex- use_dummy_dataset (bool, optional, defaults to False) — Whether to load a “dummy” variant of the dataset specified by dataset.
• label_smoothing (float, optional, defaults to 0.0) — Only relevant if return_loss is set to True. Controls the epsilon parameter value for label smoothing in the loss calculation. If set to 0, no label smoothing is performed.
• do_marginalize (bool, optional, defaults to False) — If True, the logits are marginalized over all documents by making use of torch.nn.functional.log_softmax.
• reduce_loss (bool, optional, defaults to False) — Whether or not to reduce the NLL loss using the torch.Tensor.sum operation.
• do_deduplication (bool, optional, defaults to True) — Whether or not to deduplicate the generations from different context documents for a given input. Has to be set to False if used while training with distributed backend.
• exclude_bos_score (bool, optional, defaults to False) — Whether or not to disregard the BOS token when computing the loss.
• output_retrieved(bool, optional, defaults to False) — If set to True, retrieved_doc_embeds, retrieved_doc_ids, context_input_ids and context_attention_mask are returned. See returned tensors for more detail.
• use_cache (bool, optional, defaults to True) — Whether or not the model should return the last key/values attentions (not used by all models).
• forced_eos_token_id (int, optional) — The id of the token to force as the last generated token when max_length is reached. Usually set to eos_token_id.

RagConfig stores the configuration of a RagModel. Configuration objects inherit from PretrainedConfig and can be used to control the model outputs. Read the documentation from PretrainedConfig for more information.

#### from_question_encoder_generator_configs

< >

( question_encoder_config: PretrainedConfig generator_config: PretrainedConfig **kwargs ) EncoderDecoderConfig

An instance of a configuration object

Instantiate a EncoderDecoderConfig (or a derived class) from a pre-trained encoder model configuration and decoder model configuration.

#### to_dict

< >

( ) Dict[str, any]

Returns

Dict[str, any]

Dictionary of all the attributes that make up this configuration instance,

Serializes this instance to a Python dictionary. Override the default to_dict().

## RagTokenizer

### class transformers.RagTokenizer

< >

( question_encoder generator )

#### as_target_tokenizer

< >

( )

Temporarily sets the tokenizer for encoding the targets. Useful for tokenizer associated to sequence-to-sequence models that need a slightly different processing for the labels.

## Rag specific outputs

### class transformers.models.rag.modeling_rag.RetrievAugLMMarginOutput

< >

( loss: typing.Optional[torch.FloatTensor] = None logits: FloatTensor = None doc_scores: FloatTensor = None past_key_values: typing.Optional[typing.List[torch.FloatTensor]] = None retrieved_doc_embeds: typing.Optional[torch.FloatTensor] = None retrieved_doc_ids: typing.Optional[torch.LongTensor] = None context_input_ids: typing.Optional[torch.LongTensor] = None context_attention_mask: typing.Optional[torch.LongTensor] = None question_encoder_last_hidden_state: typing.Optional[torch.FloatTensor] = None question_enc_hidden_states: typing.Optional[typing.Tuple[torch.FloatTensor]] = None question_enc_attentions: typing.Optional[typing.Tuple[torch.FloatTensor]] = None generator_enc_last_hidden_state: typing.Optional[torch.FloatTensor] = None generator_enc_hidden_states: typing.Optional[typing.Tuple[torch.FloatTensor]] = None generator_enc_attentions: typing.Optional[typing.Tuple[torch.FloatTensor]] = None generator_dec_hidden_states: typing.Optional[typing.Tuple[torch.FloatTensor]] = None generator_dec_attentions: typing.Optional[typing.Tuple[torch.FloatTensor]] = None generator_cross_attentions: typing.Optional[typing.Tuple[torch.FloatTensor]] = None )

Parameters

• loss (torch.FloatTensor of shape (1,), optional, returned when labels is provided) — Language modeling loss.
• logits (torch.FloatTensor of shape (batch_size, sequence_length, config.vocab_size)) — Prediction scores of the language modeling head. The score is possibly marginalized over all documents for each vocabulary token.
• doc_scores (torch.FloatTensor of shape (batch_size, config.n_docs)) — Score between each retrieved document embeddings (see retrieved_doc_embeds) and question_encoder_last_hidden_state.
• past_key_values (List[torch.FloatTensor], optional, returned when use_cache=True is passed or when config.use_cache=True) — List of torch.FloatTensor of length config.n_layers, with each tensor of shape (2, batch_size, num_heads, sequence_length, embed_size_per_head)).

Contains precomputed hidden-states (key and values in the attention blocks) of the decoder that can be used (see past_key_values input) to speed up sequential decoding.

• retrieved_doc_embeds (torch.FloatTensor of shape (batch_size, config.n_docs, hidden_size), optional, returned when output_retrieved=True) — Embedded documents retrieved by the retriever. Is used with question_encoder_last_hidden_state to compute the doc_scores.
• retrieved_doc_ids (torch.LongTensor of shape (batch_size, config.n_docs), optional, returned when output_retrieved=True) — The indexes of the embedded documents retrieved by the retriever.
• context_input_ids (torch.LongTensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Input ids post-processed from the retrieved documents and the question encoder input_ids by the retriever.
• context_attention_mask (torch.LongTensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Attention mask post-processed from the retrieved documents and the question encoder input_ids by the retriever.
• question_encoder_last_hidden_state (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) — Sequence of hidden states at the output of the last layer of the question encoder pooled output of the model.
• question_enc_hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the question encoder at the output of each layer plus the initial embedding outputs.

• question_enc_attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the question encoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_enc_last_hidden_state (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) — Sequence of hidden-states at the output of the last layer of the generator encoder of the model.
• generator_enc_hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the generator encoder at the output of each layer plus the initial embedding outputs.

• generator_enc_attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the generator encoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_dec_hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the generator decoder at the output of each layer plus the initial embedding outputs.

• generator_dec_attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the generator decoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_cross_attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Cross-attentions weights of the generator decoder, after the attention softmax, used to compute the weighted average in the cross-attention heads.

Base class for retriever augmented marginalized models outputs.

### class transformers.models.rag.modeling_rag.RetrievAugLMOutput

< >

( logits: FloatTensor = None doc_scores: FloatTensor = None past_key_values: typing.Optional[typing.List[torch.FloatTensor]] = None retrieved_doc_embeds: typing.Optional[torch.FloatTensor] = None retrieved_doc_ids: typing.Optional[torch.LongTensor] = None context_input_ids: typing.Optional[torch.LongTensor] = None context_attention_mask: typing.Optional[torch.LongTensor] = None question_encoder_last_hidden_state: typing.Optional[torch.FloatTensor] = None question_enc_hidden_states: typing.Optional[typing.Tuple[torch.FloatTensor]] = None question_enc_attentions: typing.Optional[typing.Tuple[torch.FloatTensor]] = None generator_enc_last_hidden_state: typing.Optional[torch.FloatTensor] = None generator_enc_hidden_states: typing.Optional[typing.Tuple[torch.FloatTensor]] = None generator_enc_attentions: typing.Optional[typing.Tuple[torch.FloatTensor]] = None generator_dec_hidden_states: typing.Optional[typing.Tuple[torch.FloatTensor]] = None generator_dec_attentions: typing.Optional[typing.Tuple[torch.FloatTensor]] = None generator_cross_attentions: typing.Optional[typing.Tuple[torch.FloatTensor]] = None )

Parameters

• logits (torch.FloatTensor of shape (batch_size, sequence_length, config.vocab_size)) — Prediction scores of the language modeling head. The score is possibly marginalized over all documents for each vocabulary token.
• doc_scores (torch.FloatTensor of shape (batch_size, config.n_docs)) — Score between each retrieved document embeddings (see retrieved_doc_embeds) and question_encoder_last_hidden_state.
• past_key_values (List[torch.FloatTensor], optional, returned when use_cache=True is passed or when config.use_cache=True) — List of torch.FloatTensor of length config.n_layers, with each tensor of shape (2, batch_size, num_heads, sequence_length, embed_size_per_head)).

Contains precomputed hidden-states (key and values in the attention blocks) of the decoder that can be used (see past_key_values input) to speed up sequential decoding.

• retrieved_doc_embeds (torch.FloatTensor of shape (batch_size, config.n_docs, hidden_size), optional, returned when output_retrieved=True) — Embedded documents retrieved by the retriever. Is used with question_encoder_last_hidden_state to compute the doc_scores.
• retrieved_doc_ids (torch.LongTensor of shape (batch_size, config.n_docs), optional, returned when output_retrieved=True) — The indexes of the embedded documents retrieved by the retriever.
• context_input_ids (torch.LongTensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Input ids post-processed from the retrieved documents and the question encoder input_ids by the retriever.
• context_attention_mask (torch.LongTensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Attention mask post-processed from the retrieved documents and the question encoder input_ids by the retriever.
• question_encoder_last_hidden_state (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) — Sequence of hidden states at the output of the last layer of the question encoder pooled output of the model.
• question_enc_hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the question encoder at the output of each layer plus the initial embedding outputs.

• question_enc_attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the question encoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_enc_last_hidden_state (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) — Sequence of hidden-states at the output of the last layer of the generator encoder of the model.
• generator_enc_hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the generator encoder at the output of each layer plus the initial embedding outputs.

• generator_enc_attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the generator encoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_dec_hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the generator decoder at the output of each layer plus the initial embedding outputs.

• generator_dec_attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the generator decoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_cross_attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Cross-attentions weights of the generator decoder, after the attention softmax, used to compute the weighted average in the cross-attention heads.

## RagRetriever

### class transformers.RagRetriever

< >

( config question_encoder_tokenizer generator_tokenizer index = None init_retrieval = True )

Parameters

• config (RagConfig) — The configuration of the RAG model this Retriever is used with. Contains parameters indicating which Index to build. You can load your own custom dataset with config.index_name="custom" or use a canonical one (default) from the datasets library with config.index_name="wiki_dpr" for example.
• question_encoder_tokenizer (PreTrainedTokenizer) — The tokenizer that was used to tokenize the question. It is used to decode the question and then use the generator_tokenizer.
• generator_tokenizer (PreTrainedTokenizer) — The tokenizer used for the generator part of the RagModel.
• index (Index optional, defaults to the one defined by the configuration) — If specified, use this index instead of the one built using the configuration

Retriever used to get documents from vector queries. It retrieves the documents embeddings as well as the documents contents, and it formats them to be used with a RagModel.

Examples:

>>> # To load the default "wiki_dpr" dataset with 21M passages from wikipedia (index name is 'compressed' or 'exact')
>>> from transformers import RagRetriever

>>> retriever = RagRetriever.from_pretrained(
... )

>>> # To load your own indexed dataset built with the datasets library. More info on how to build the indexed dataset in examples/rag/use_own_knowledge_dataset.py
>>> from transformers import RagRetriever

>>> dataset = (
...     ...
>>> )  # dataset must be a datasets.Datasets object with columns "title", "text" and "embeddings", and it must have a faiss index

>>> from transformers import RagRetriever

>>> dataset_path = "path/to/my/dataset"  # dataset saved via *dataset.save_to_disk(...)*
>>> index_path = "path/to/my/index.faiss"  # faiss index saved via *dataset.get_index("embeddings").save(...)*
>>> retriever = RagRetriever.from_pretrained(
...     index_name="custom",
...     passages_path=dataset_path,
...     index_path=index_path,
... )

>>> # To load the legacy index built originally for Rag's paper
>>> from transformers import RagRetriever

>>> retriever = RagRetriever.from_pretrained("facebook/dpr-ctx_encoder-single-nq-base", index_name="legacy")

#### init_retrieval

< >

( )

Retriever initialization function. It loads the index into memory.

#### postprocess_docs

< >

( docs input_strings prefix n_docs return_tensors = None ) tuple(tensors)

Parameters

• docs (dict) — Retrieved documents.
• input_strings (str) — Input strings decoded by preprocess_query.
• prefix (str) — Prefix added at the beginning of each input, typically used with T5-based models.

Returns

tuple(tensors)

a tuple consisting of two elements: contextualized input_ids and a compatible attention_mask.

Postprocessing retrieved docs and combining them with input_strings.

#### retrieve

< >

( question_hidden_states: ndarray n_docs: int ) Tuple[np.ndarray, np.ndarray, List[dict]]

Parameters

• question_hidden_states (np.ndarray of shape (batch_size, vector_size)) — A batch of query vectors to retrieve with.
• n_docs (int) — The number of docs retrieved per query.

Returns

Tuple[np.ndarray, np.ndarray, List[dict]]

A tuple with the following objects:

• retrieved_doc_embeds (np.ndarray of shape (batch_size, n_docs, dim)) — The retrieval embeddings of the retrieved docs per query.
• doc_ids (np.ndarray of shape (batch_size, n_docs)) — The ids of the documents in the index
• doc_dicts (List[dict]): The retrieved_doc_embeds examples per query.

Retrieves documents for specified question_hidden_states.

## RagModel

### class transformers.RagModel

< >

( config: typing.Optional[transformers.configuration_utils.PretrainedConfig] = None question_encoder: typing.Optional[transformers.modeling_utils.PreTrainedModel] = None generator: typing.Optional[transformers.modeling_utils.PreTrainedModel] = None retriever: typing.Optional = None **kwargs )

Parameters

• config (RagConfig) — Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.
• question_encoder (PreTrainedModel) — An encoder model compatible with the faiss index encapsulated by the retriever.
• generator (PreTrainedModel) — A seq2seq model used as the generator in the RAG architecture.
• retriever (RagRetriever) — A retriever class encapsulating a faiss index queried to obtain context documents for current inputs.

The RagModel forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

RAG is a seq2seq model which encapsulates two core components: a question encoder and a generator. During a forward pass, we encode the input with the question encoder and pass it to the retriever to extract relevant context documents. The documents are then prepended to the input. Such contextualized inputs is passed to the generator.

The question encoder can be any autoencoding model, preferably DPRQuestionEncoder, and the generator can be any seq2seq model, preferably BartForConditionalGeneration.

The model can be initialized with a RagRetriever for end-to-end generation or used in combination with the outputs of a retriever in multiple steps---see examples for more details. The model is compatible any autoencoding model as the question_encoder and any seq2seq model with language model head as the generator. It has been tested with DPRQuestionEncoder as the question_encoder and BartForConditionalGeneration or T5ForConditionalGeneration as the generator.

This model inherits from PreTrainedModel. Check the superclass documentation for the generic methods the library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads etc.)

This model is also a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.

#### forward

< >

( input_ids = None attention_mask = None encoder_outputs = None decoder_input_ids = None decoder_attention_mask = None past_key_values = None doc_scores = None context_input_ids = None context_attention_mask = None use_cache = None output_attentions = None output_hidden_states = None output_retrieved = None n_docs = None ) transformers.models.rag.modeling_rag.RetrievAugLMOutput or tuple(torch.FloatTensor)

Parameters

• input_ids (torch.LongTensor of shape (batch_size, sequence_length)) — Indices of input sequence tokens in the vocabulary. RagConfig, used to initialize the model, specifies which generator to use, it also specifies a compatible generator tokenizer. Use that tokenizer class to obtain the indices.

What are input IDs?

• attention_mask (torch.Tensor of shape (batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

• 1 for tokens that are not masked,
• 0 for tokens that are masked.

• encoder_outputs (tuple(tuple(torch.FloatTensor), optional) — Tuple consists of (generator_enc_last_hidden_state, optional: generator_enc_hidden_states, optional: generator_enc_attentions). generator_enc_last_hidden_state of shape (batch_size, n_docs * sequence_length, hidden_size) is a sequence of hidden-states at the output of the last layer of the generator’s encoder.

Used by the (RagModel) model during decoding.

• decoder_input_ids (torch.LongTensor of shape (batch_size, target_sequence_length), optional) — Provide for generation tasks. None by default, construct as per instructions for the generator model you’re using with your RAG instance.
• decoder_attention_mask (torch.BoolTensor of shape (batch_size, target_sequence_length), optional) — Default behavior: generate a tensor that ignores pad tokens in decoder_input_ids. Causal mask will also be used by default.
• past_key_values (tuple(tuple(torch.FloatTensor))) — Tuple consists of two elements: encoder_outputs of the RAG model (see encoder_outputs) and past_key_values of the underlying generator. Can be used to speed up decoding. past_key_values are used in the (RagTokenForGeneration) model during decoding.
• doc_scores (torch.FloatTensor of shape (batch_size, config.n_docs)) — Score between each retrieved document embeddings (see retrieved_doc_embeds) and question_encoder_last_hidden_state. If the model has is not initialized with a retriever doc_scores has to be provided to the forward pass. doc_scores can be computed via question_encoder_last_hidden_state and retrieved_doc_embeds, see examples for more information.
• context_input_ids (torch.LongTensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Input IDs post-processed from the retrieved documents and the question encoder input_ids by the retriever.

If the model has is not initialized with a retriever context_input_ids has to be provided to the forward pass. context_input_ids are returned by __call__() context_attention_mask (torch.LongTensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True): Attention mask post-processed from the retrieved documents and the question encoder input_ids by the retriever.

If the model has is not initialized with a retriever context_attention_mask has to be provided to the forward pass. context_attention_mask are returned by __call__()

• use_cache (bool, optional, defaults to True) — If set to True, past_key_values key value states are returned and can be used to speed up decoding (see past_key_values).
• output_attentions (bool, optional) — Whether or not to return the attentions tensors of all attention layers. See attentions under returned tensors for more detail.
• output_hidden_states (bool, optional) — Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for more detail.
• output_retrieved(bool, optional) — Whether or not to return the retrieved_doc_embeds, retrieved_doc_ids, context_input_ids and context_attention_mask. See returned tensors for more detail.
• n_docs (int, optional, defaults to config.n_docs“) — Number of documents to retrieve and/or number of documents for which to generate an answer.

Returns

transformers.models.rag.modeling_rag.RetrievAugLMOutput or tuple(torch.FloatTensor)

A transformers.models.rag.modeling_rag.RetrievAugLMOutput or a tuple of torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (RagConfig) and inputs.

• logits (torch.FloatTensor of shape (batch_size, sequence_length, config.vocab_size)) — Prediction scores of the language modeling head. The score is possibly marginalized over all documents for each vocabulary token.

• doc_scores (torch.FloatTensor of shape (batch_size, config.n_docs)) — Score between each retrieved document embeddings (see retrieved_doc_embeds) and question_encoder_last_hidden_state.

• past_key_values (List[torch.FloatTensor], optional, returned when use_cache=True is passed or when config.use_cache=True) — List of torch.FloatTensor of length config.n_layers, with each tensor of shape (2, batch_size, num_heads, sequence_length, embed_size_per_head)).

Contains precomputed hidden-states (key and values in the attention blocks) of the decoder that can be used (see past_key_values input) to speed up sequential decoding.

• retrieved_doc_embeds (torch.FloatTensor of shape (batch_size, config.n_docs, hidden_size), optional, returned when output_retrieved=True) — Embedded documents retrieved by the retriever. Is used with question_encoder_last_hidden_state to compute the doc_scores.

• retrieved_doc_ids (torch.LongTensor of shape (batch_size, config.n_docs), optional, returned when output_retrieved=True) — The indexes of the embedded documents retrieved by the retriever.

• context_input_ids (torch.LongTensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Input ids post-processed from the retrieved documents and the question encoder input_ids by the retriever.

• context_attention_mask (torch.LongTensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Attention mask post-processed from the retrieved documents and the question encoder input_ids by the retriever.

• question_encoder_last_hidden_state (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) — Sequence of hidden states at the output of the last layer of the question encoder pooled output of the model.

• question_enc_hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the question encoder at the output of each layer plus the initial embedding outputs.

• question_enc_attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the question encoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_enc_last_hidden_state (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) — Sequence of hidden-states at the output of the last layer of the generator encoder of the model.

• generator_enc_hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the generator encoder at the output of each layer plus the initial embedding outputs.

• generator_enc_attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the generator encoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_dec_hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the generator decoder at the output of each layer plus the initial embedding outputs.

• generator_dec_attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the generator decoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_cross_attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Cross-attentions weights of the generator decoder, after the attention softmax, used to compute the weighted average in the cross-attention heads.

The RagModel forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example:

>>> from transformers import RagTokenizer, RagRetriever, RagModel
>>> import torch

>>> retriever = RagRetriever.from_pretrained(
... )
>>> # initialize with RagRetriever to do everything in one forward call

>>> inputs = tokenizer("How many people live in Paris?", return_tensors="pt")
>>> outputs = model(input_ids=inputs["input_ids"])

## RagSequenceForGeneration

### class transformers.RagSequenceForGeneration

< >

( config: typing.Optional[transformers.configuration_utils.PretrainedConfig] = None question_encoder: typing.Optional[transformers.modeling_utils.PreTrainedModel] = None generator: typing.Optional[transformers.modeling_utils.PreTrainedModel] = None retriever: typing.Optional = None **kwargs )

Parameters

• config (RagConfig) — Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.
• question_encoder (PreTrainedModel) — An encoder model compatible with the faiss index encapsulated by the retriever.
• generator (PreTrainedModel) — A seq2seq model used as the generator in the RAG architecture.
• retriever (RagRetriever) — A retriever class encapsulating a faiss index queried to obtain context documents for current inputs.

The RagSequenceForGeneration forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

A RAG-sequence model implementation. It performs RAG-sequence specific marginalization in the forward pass.

RAG is a seq2seq model which encapsulates two core components: a question encoder and a generator. During a forward pass, we encode the input with the question encoder and pass it to the retriever to extract relevant context documents. The documents are then prepended to the input. Such contextualized inputs is passed to the generator.

The question encoder can be any autoencoding model, preferably DPRQuestionEncoder, and the generator can be any seq2seq model, preferably BartForConditionalGeneration.

The model can be initialized with a RagRetriever for end-to-end generation or used in combination with the outputs of a retriever in multiple steps---see examples for more details. The model is compatible any autoencoding model as the question_encoder and any seq2seq model with language model head as the generator. It has been tested with DPRQuestionEncoder as the question_encoder and BartForConditionalGeneration or T5ForConditionalGeneration as the generator.

This model inherits from PreTrainedModel. Check the superclass documentation for the generic methods the library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads etc.)

This model is also a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.

#### forward

< >

( input_ids = None attention_mask = None encoder_outputs = None decoder_input_ids = None decoder_attention_mask = None past_key_values = None context_input_ids = None context_attention_mask = None doc_scores = None use_cache = None output_attentions = None output_hidden_states = None output_retrieved = None exclude_bos_score = None reduce_loss = None labels = None n_docs = None **kwargs ) transformers.models.rag.modeling_rag.RetrievAugLMMarginOutput or tuple(torch.FloatTensor)

Parameters

• input_ids (torch.LongTensor of shape (batch_size, sequence_length)) — Indices of input sequence tokens in the vocabulary. RagConfig, used to initialize the model, specifies which generator to use, it also specifies a compatible generator tokenizer. Use that tokenizer class to obtain the indices.

What are input IDs?

• attention_mask (torch.Tensor of shape (batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

• 1 for tokens that are not masked,
• 0 for tokens that are masked.

• encoder_outputs (tuple(tuple(torch.FloatTensor), optional) — Tuple consists of (generator_enc_last_hidden_state, optional: generator_enc_hidden_states, optional: generator_enc_attentions). generator_enc_last_hidden_state of shape (batch_size, n_docs * sequence_length, hidden_size) is a sequence of hidden-states at the output of the last layer of the generator’s encoder.

Used by the (RagModel) model during decoding.

• decoder_input_ids (torch.LongTensor of shape (batch_size, target_sequence_length), optional) — Provide for generation tasks. None by default, construct as per instructions for the generator model you’re using with your RAG instance.
• decoder_attention_mask (torch.BoolTensor of shape (batch_size, target_sequence_length), optional) — Default behavior: generate a tensor that ignores pad tokens in decoder_input_ids. Causal mask will also be used by default.
• past_key_values (tuple(tuple(torch.FloatTensor))) — Tuple consists of two elements: encoder_outputs of the RAG model (see encoder_outputs) and past_key_values of the underlying generator. Can be used to speed up decoding. past_key_values are used in the (RagTokenForGeneration) model during decoding.
• doc_scores (torch.FloatTensor of shape (batch_size, config.n_docs)) — Score between each retrieved document embeddings (see retrieved_doc_embeds) and question_encoder_last_hidden_state. If the model has is not initialized with a retriever doc_scores has to be provided to the forward pass. doc_scores can be computed via question_encoder_last_hidden_state and retrieved_doc_embeds, see examples for more information.
• context_input_ids (torch.LongTensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Input IDs post-processed from the retrieved documents and the question encoder input_ids by the retriever.

If the model has is not initialized with a retriever context_input_ids has to be provided to the forward pass. context_input_ids are returned by __call__() context_attention_mask (torch.LongTensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True): Attention mask post-processed from the retrieved documents and the question encoder input_ids by the retriever.

If the model has is not initialized with a retriever context_attention_mask has to be provided to the forward pass. context_attention_mask are returned by __call__()

• use_cache (bool, optional, defaults to True) — If set to True, past_key_values key value states are returned and can be used to speed up decoding (see past_key_values).
• output_attentions (bool, optional) — Whether or not to return the attentions tensors of all attention layers. See attentions under returned tensors for more detail.
• output_hidden_states (bool, optional) — Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for more detail.
• output_retrieved(bool, optional) — Whether or not to return the retrieved_doc_embeds, retrieved_doc_ids, context_input_ids and context_attention_mask. See returned tensors for more detail.
• n_docs (int, optional, defaults to config.n_docs“) — Number of documents to retrieve and/or number of documents for which to generate an answer.
• exclude_bos_score (bool, optional) — Only relevant if labels is passed. If True, the score of the BOS token is disregarded when computing the loss.
• reduce_loss (bool, optional) — Only relevant if labels is passed. If True, the NLL loss is reduced using the torch.Tensor.sum operation.
• kwargs (Dict[str, any], optional, defaults to {}) — Legacy dictionary, which is required so that model can use generate() function.

Returns

transformers.models.rag.modeling_rag.RetrievAugLMMarginOutput or tuple(torch.FloatTensor)

A transformers.models.rag.modeling_rag.RetrievAugLMMarginOutput or a tuple of torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (RagConfig) and inputs.

• loss (torch.FloatTensor of shape (1,), optional, returned when labels is provided) — Language modeling loss.

• logits (torch.FloatTensor of shape (batch_size, sequence_length, config.vocab_size)) — Prediction scores of the language modeling head. The score is possibly marginalized over all documents for each vocabulary token.

• doc_scores (torch.FloatTensor of shape (batch_size, config.n_docs)) — Score between each retrieved document embeddings (see retrieved_doc_embeds) and question_encoder_last_hidden_state.

• past_key_values (List[torch.FloatTensor], optional, returned when use_cache=True is passed or when config.use_cache=True) — List of torch.FloatTensor of length config.n_layers, with each tensor of shape (2, batch_size, num_heads, sequence_length, embed_size_per_head)).

Contains precomputed hidden-states (key and values in the attention blocks) of the decoder that can be used (see past_key_values input) to speed up sequential decoding.

• retrieved_doc_embeds (torch.FloatTensor of shape (batch_size, config.n_docs, hidden_size), optional, returned when output_retrieved=True) — Embedded documents retrieved by the retriever. Is used with question_encoder_last_hidden_state to compute the doc_scores.

• retrieved_doc_ids (torch.LongTensor of shape (batch_size, config.n_docs), optional, returned when output_retrieved=True) — The indexes of the embedded documents retrieved by the retriever.

• context_input_ids (torch.LongTensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Input ids post-processed from the retrieved documents and the question encoder input_ids by the retriever.

• context_attention_mask (torch.LongTensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Attention mask post-processed from the retrieved documents and the question encoder input_ids by the retriever.

• question_encoder_last_hidden_state (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) — Sequence of hidden states at the output of the last layer of the question encoder pooled output of the model.

• question_enc_hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the question encoder at the output of each layer plus the initial embedding outputs.

• question_enc_attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the question encoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_enc_last_hidden_state (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) — Sequence of hidden-states at the output of the last layer of the generator encoder of the model.

• generator_enc_hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the generator encoder at the output of each layer plus the initial embedding outputs.

• generator_enc_attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the generator encoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_dec_hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the generator decoder at the output of each layer plus the initial embedding outputs.

• generator_dec_attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the generator decoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_cross_attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Cross-attentions weights of the generator decoder, after the attention softmax, used to compute the weighted average in the cross-attention heads.

The RagSequenceForGeneration forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example:

>>> from transformers import RagTokenizer, RagRetriever, RagSequenceForGeneration
>>> import torch

>>> retriever = RagRetriever.from_pretrained(
... )
>>> # initialize with RagRetriever to do everything in one forward call

>>> inputs = tokenizer("How many people live in Paris?", return_tensors="pt")
>>> with tokenizer.as_target_tokenizer():
...     targets = tokenizer("In Paris, there are 10 million people.", return_tensors="pt")
>>> input_ids = inputs["input_ids"]
>>> labels = targets["input_ids"]
>>> outputs = model(input_ids=input_ids, labels=labels)

>>> # or use retriever separately
>>> # 1. Encode
>>> question_hidden_states = model.question_encoder(input_ids)[0]
>>> # 2. Retrieve
>>> docs_dict = retriever(input_ids.numpy(), question_hidden_states.detach().numpy(), return_tensors="pt")
>>> doc_scores = torch.bmm(
...     question_hidden_states.unsqueeze(1), docs_dict["retrieved_doc_embeds"].float().transpose(1, 2)
>>> ).squeeze(1)
>>> # 3. Forward to generator
>>> outputs = model(
...     context_input_ids=docs_dict["context_input_ids"],
...     doc_scores=doc_scores,
...     decoder_input_ids=labels,
... )

#### generate

< >

( input_ids: typing.Optional[torch.LongTensor] = None attention_mask: typing.Optional[torch.LongTensor] = None context_input_ids = None context_attention_mask = None doc_scores = None do_deduplication = None num_return_sequences = None num_beams = None n_docs = None **model_kwargs ) torch.LongTensor of shape (batch_size * num_return_sequences, sequence_length)

Parameters

• input_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) — The sequence used as a prompt for the generation. If input_ids is not passed, then context_input_ids has to be provided.
• attention_mask (torch.Tensor of shape (batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

• 1 for tokens that are not masked,
• 0 for tokens that are masked.

• context_input_ids (torch.LongTensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Input IDs post-processed from the retrieved documents and the question encoder input_ids by the retriever.
• context_attention_mask (torch.LongTensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Attention mask post-processed from the retrieved documents and the question encoder input_ids by the retriever.

If the model is not initialized with a retriever or input_ids is not given, context_input_ids and context_attention_mask have to be provided to the forward pass. They are returned by __call__()

• doc_scores (torch.FloatTensor of shape (batch_size, config.n_docs)) — Score between each retrieved document embeddings (see retrieved_doc_embeds) and question_encoder_last_hidden_state.

If the model is not initialized with a retriever or input_ids is not given, doc_scores has to be provided to the forward pass. doc_scores are returned by __call__()

• do_deduplication (bool, optional) — Whether or not to deduplicate the generations from different context documents for a given input. Has to be set to False if used while training with distributed backend.
• num_return_sequences(int, optional, defaults to 1) — The number of independently computed returned sequences for each element in the batch. Note that this is not the value we pass to the generator’s [generate()](/docs/transformers/v4.17.0/en/main_classes/model#transformers.generation_utils.GenerationMixin.generate) function, where we set num_return_sequences to num_beams.
• num_beams (int, optional, defaults to 1) — Number of beams for beam search. 1 means no beam search.
• n_docs (int, optional, defaults to config.n_docs) — Number of documents to retrieve and/or number of documents for which to generate an answer. kwargs — Additional kwargs will be passed to generate().

Returns

torch.LongTensor of shape (batch_size * num_return_sequences, sequence_length)

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.

Implements RAG sequence “thorough” decoding. Read the generate() documentation for more information on how to set other generate input parameters.

## RagTokenForGeneration

### class transformers.RagTokenForGeneration

< >

( config: typing.Optional[transformers.configuration_utils.PretrainedConfig] = None question_encoder: typing.Optional[transformers.modeling_utils.PreTrainedModel] = None generator: typing.Optional[transformers.modeling_utils.PreTrainedModel] = None retriever: typing.Optional = None **kwargs )

Parameters

• config (RagConfig) — Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.
• question_encoder (PreTrainedModel) — An encoder model compatible with the faiss index encapsulated by the retriever.
• generator (PreTrainedModel) — A seq2seq model used as the generator in the RAG architecture.
• retriever (RagRetriever) — A retriever class encapsulating a faiss index queried to obtain context documents for current inputs.

The RagTokenForGeneration forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

A RAG-token model implementation. It performs RAG-token specific marginalization in the forward pass.

RAG is a seq2seq model which encapsulates two core components: a question encoder and a generator. During a forward pass, we encode the input with the question encoder and pass it to the retriever to extract relevant context documents. The documents are then prepended to the input. Such contextualized inputs is passed to the generator.

The question encoder can be any autoencoding model, preferably DPRQuestionEncoder, and the generator can be any seq2seq model, preferably BartForConditionalGeneration.

The model can be initialized with a RagRetriever for end-to-end generation or used in combination with the outputs of a retriever in multiple steps---see examples for more details. The model is compatible any autoencoding model as the question_encoder and any seq2seq model with language model head as the generator. It has been tested with DPRQuestionEncoder as the question_encoder and BartForConditionalGeneration or T5ForConditionalGeneration as the generator.

This model inherits from PreTrainedModel. Check the superclass documentation for the generic methods the library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads etc.)

This model is also a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.

#### forward

< >

( input_ids = None attention_mask = None encoder_outputs = None decoder_input_ids = None decoder_attention_mask = None past_key_values = None context_input_ids = None context_attention_mask = None doc_scores = None use_cache = None output_attentions = None output_hidden_states = None output_retrieved = None do_marginalize = None reduce_loss = None labels = None n_docs = None **kwargs ) transformers.models.rag.modeling_rag.RetrievAugLMMarginOutput or tuple(torch.FloatTensor)

Parameters

• input_ids (torch.LongTensor of shape (batch_size, sequence_length)) — Indices of input sequence tokens in the vocabulary. RagConfig, used to initialize the model, specifies which generator to use, it also specifies a compatible generator tokenizer. Use that tokenizer class to obtain the indices.

What are input IDs?

• attention_mask (torch.Tensor of shape (batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

• 1 for tokens that are not masked,
• 0 for tokens that are masked.

• encoder_outputs (tuple(tuple(torch.FloatTensor), optional) — Tuple consists of (generator_enc_last_hidden_state, optional: generator_enc_hidden_states, optional: generator_enc_attentions). generator_enc_last_hidden_state of shape (batch_size, n_docs * sequence_length, hidden_size) is a sequence of hidden-states at the output of the last layer of the generator’s encoder.

Used by the (RagModel) model during decoding.

• decoder_input_ids (torch.LongTensor of shape (batch_size, target_sequence_length), optional) — Provide for generation tasks. None by default, construct as per instructions for the generator model you’re using with your RAG instance.
• decoder_attention_mask (torch.BoolTensor of shape (batch_size, target_sequence_length), optional) — Default behavior: generate a tensor that ignores pad tokens in decoder_input_ids. Causal mask will also be used by default.
• past_key_values (tuple(tuple(torch.FloatTensor))) — Tuple consists of two elements: encoder_outputs of the RAG model (see encoder_outputs) and past_key_values of the underlying generator. Can be used to speed up decoding. past_key_values are used in the (RagTokenForGeneration) model during decoding.
• doc_scores (torch.FloatTensor of shape (batch_size, config.n_docs)) — Score between each retrieved document embeddings (see retrieved_doc_embeds) and question_encoder_last_hidden_state. If the model has is not initialized with a retriever doc_scores has to be provided to the forward pass. doc_scores can be computed via question_encoder_last_hidden_state and retrieved_doc_embeds, see examples for more information.
• context_input_ids (torch.LongTensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Input IDs post-processed from the retrieved documents and the question encoder input_ids by the retriever.

If the model has is not initialized with a retriever context_input_ids has to be provided to the forward pass. context_input_ids are returned by __call__() context_attention_mask (torch.LongTensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True): Attention mask post-processed from the retrieved documents and the question encoder input_ids by the retriever.

If the model has is not initialized with a retriever context_attention_mask has to be provided to the forward pass. context_attention_mask are returned by __call__()

• use_cache (bool, optional, defaults to True) — If set to True, past_key_values key value states are returned and can be used to speed up decoding (see past_key_values).
• output_attentions (bool, optional) — Whether or not to return the attentions tensors of all attention layers. See attentions under returned tensors for more detail.
• output_hidden_states (bool, optional) — Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for more detail.
• output_retrieved(bool, optional) — Whether or not to return the retrieved_doc_embeds, retrieved_doc_ids, context_input_ids and context_attention_mask. See returned tensors for more detail.
• n_docs (int, optional, defaults to config.n_docs“) — Number of documents to retrieve and/or number of documents for which to generate an answer.
• do_marginalize (bool, optional) — If True, the logits are marginalized over all documents by making use of torch.nn.functional.log_softmax.
• reduce_loss (bool, optional) — Only relevant if labels is passed. If True, the NLL loss is reduced using the torch.Tensor.sum operation.
• kwargs (Dict[str, any], optional, defaults to {}) — Legacy dictionary, which is required so that model can use generate() function.

Returns

transformers.models.rag.modeling_rag.RetrievAugLMMarginOutput or tuple(torch.FloatTensor)

A transformers.models.rag.modeling_rag.RetrievAugLMMarginOutput or a tuple of torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (RagConfig) and inputs.

• loss (torch.FloatTensor of shape (1,), optional, returned when labels is provided) — Language modeling loss.

• logits (torch.FloatTensor of shape (batch_size, sequence_length, config.vocab_size)) — Prediction scores of the language modeling head. The score is possibly marginalized over all documents for each vocabulary token.

• doc_scores (torch.FloatTensor of shape (batch_size, config.n_docs)) — Score between each retrieved document embeddings (see retrieved_doc_embeds) and question_encoder_last_hidden_state.

• past_key_values (List[torch.FloatTensor], optional, returned when use_cache=True is passed or when config.use_cache=True) — List of torch.FloatTensor of length config.n_layers, with each tensor of shape (2, batch_size, num_heads, sequence_length, embed_size_per_head)).

Contains precomputed hidden-states (key and values in the attention blocks) of the decoder that can be used (see past_key_values input) to speed up sequential decoding.

• retrieved_doc_embeds (torch.FloatTensor of shape (batch_size, config.n_docs, hidden_size), optional, returned when output_retrieved=True) — Embedded documents retrieved by the retriever. Is used with question_encoder_last_hidden_state to compute the doc_scores.

• retrieved_doc_ids (torch.LongTensor of shape (batch_size, config.n_docs), optional, returned when output_retrieved=True) — The indexes of the embedded documents retrieved by the retriever.

• context_input_ids (torch.LongTensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Input ids post-processed from the retrieved documents and the question encoder input_ids by the retriever.

• context_attention_mask (torch.LongTensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Attention mask post-processed from the retrieved documents and the question encoder input_ids by the retriever.

• question_encoder_last_hidden_state (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) — Sequence of hidden states at the output of the last layer of the question encoder pooled output of the model.

• question_enc_hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the question encoder at the output of each layer plus the initial embedding outputs.

• question_enc_attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the question encoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_enc_last_hidden_state (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) — Sequence of hidden-states at the output of the last layer of the generator encoder of the model.

• generator_enc_hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the generator encoder at the output of each layer plus the initial embedding outputs.

• generator_enc_attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the generator encoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_dec_hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the generator decoder at the output of each layer plus the initial embedding outputs.

• generator_dec_attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the generator decoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_cross_attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Cross-attentions weights of the generator decoder, after the attention softmax, used to compute the weighted average in the cross-attention heads.

The RagTokenForGeneration forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example:

>>> from transformers import RagTokenizer, RagRetriever, RagTokenForGeneration
>>> import torch

>>> retriever = RagRetriever.from_pretrained(
... )
>>> # initialize with RagRetriever to do everything in one forward call

>>> inputs = tokenizer("How many people live in Paris?", return_tensors="pt")
>>> with tokenizer.as_target_tokenizer():
...     targets = tokenizer("In Paris, there are 10 million people.", return_tensors="pt")
>>> input_ids = inputs["input_ids"]
>>> labels = targets["input_ids"]
>>> outputs = model(input_ids=input_ids, labels=labels)

>>> # or use retriever separately
>>> # 1. Encode
>>> question_hidden_states = model.question_encoder(input_ids)[0]
>>> # 2. Retrieve
>>> docs_dict = retriever(input_ids.numpy(), question_hidden_states.detach().numpy(), return_tensors="pt")
>>> doc_scores = torch.bmm(
...     question_hidden_states.unsqueeze(1), docs_dict["retrieved_doc_embeds"].float().transpose(1, 2)
>>> ).squeeze(1)
>>> # 3. Forward to generator
>>> outputs = model(
...     context_input_ids=docs_dict["context_input_ids"],
...     doc_scores=doc_scores,
...     decoder_input_ids=labels,
... )

>>> # or directly generate
>>> generated = model.generate(
...     context_input_ids=docs_dict["context_input_ids"],
...     doc_scores=doc_scores,
... )
>>> generated_string = tokenizer.batch_decode(generated, skip_special_tokens=True)

#### generate

< >

( input_ids: typing.Optional[torch.LongTensor] = None attention_mask: typing.Optional[torch.LongTensor] = None context_input_ids = None context_attention_mask = None doc_scores = None max_length = None min_length = None early_stopping = None use_cache = None num_beams = None num_beam_groups = None diversity_penalty = None bos_token_id = None pad_token_id = None eos_token_id = None length_penalty = None no_repeat_ngram_size = None encoder_no_repeat_ngram_size = None repetition_penalty = None bad_words_ids = None num_return_sequences = None decoder_start_token_id = None n_docs = None prefix_allowed_tokens_fn: typing.Callable[[int, torch.Tensor], typing.List[int]] = None logits_processor: typing.Optional[transformers.generation_logits_process.LogitsProcessorList] = [] stopping_criteria: typing.Optional[transformers.generation_stopping_criteria.StoppingCriteriaList] = [] forced_bos_token_id: typing.Optional[int] = None forced_eos_token_id: typing.Optional[int] = None remove_invalid_values: typing.Optional[bool] = None **model_kwargs ) torch.LongTensor of shape (batch_size * num_return_sequences, sequence_length)

Parameters

• input_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) — The sequence used as a prompt for the generation. If input_ids is not passed, then context_input_ids has to be provided.
• attention_mask (torch.Tensor of shape (batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

• 1 for tokens that are not masked,
• 0 for tokens that are masked.

• context_input_ids (torch.LongTensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Input IDs post-processed from the retrieved documents and the question encoder input_ids by the retriever.

If the model has is not initialized with a retriever, context_input_ids has to be provided to the forward pass. context_input_ids are returned by __call__()

• context_attention_mask (torch.LongTensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Attention mask post-processed from the retrieved documents and the question encoder input_ids by the retriever.

If the model has is not initialized with a retriever, context_input_ids has to be provided to the forward pass. context_input_ids are returned by __call__()

• doc_scores (torch.FloatTensor of shape (batch_size, config.n_docs)) — Score between each retrieved document embeddings (see retrieved_doc_embeds) and question_encoder_last_hidden_state.

If the model has is not initialized with a retriever, context_input_ids has to be provided to the forward pass. context_input_ids are returned by __call__()

• max_length (int, optional, defaults to 20) — The maximum length of the sequence to be generated.
• min_length (int, optional, defaults to 10) — The minimum length of the sequence to be generated.
• early_stopping (bool, optional, defaults to False) — Whether or not to stop the beam search when at least num_beams sentences are finished per batch or not. 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.
• 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 (int, optional) — The id of the end-of-sequence token.
• length_penalty (float, optional, defaults to 1.0) — Exponential penalty to the length. 1.0 means no penalty.

Set to values < 1.0 in order to encourage the model to generate shorter sequences, to a value > 1.0 in order to encourage the model to produce longer sequences.

• no_repeat_ngram_size (int, optional, defaults to 0) — If set to int > 0, all ngrams of that size can only occur once.
• 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.
• bad_words_ids(List[int], optional) — List of token ids that are not allowed to be generated. In order to get the tokens of the words that should not appear in the generated text, use tokenizer.encode(bad_word, add_prefix_space=True).
• 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.
• 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.
• num_return_sequences(int, optional, defaults to 1) — The number of independently computed returned sequences for each element in the batch. Note that this is not the value we pass to the generator’s [generate()](/docs/transformers/v4.17.0/en/main_classes/model#transformers.generation_utils.GenerationMixin.generate) function, where we set num_return_sequencestonum_beams. decoder_start_token_id (int, optional): If an encoder-decoder model starts decoding with a different token than bos, the id of that token.
• n_docs (int, optional, defaults to config.n_docs) — Number of documents to retrieve and/or number of documents for which to generate an answer. 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 inputs_ids and the batch ID batch_id. It has to return a list with the allowed tokens for the next generation step conditioned on the previously generated tokens inputs_ids and the batch ID batch_id. This argument is useful for constrained generation conditioned on the prefix, as described in Autoregressive Entity Retrieval.
• logits_processor (LogitsProcessorList, optional) — Custom logits processors that complement the default logits processors built from arguments and a model’s config. If a logit processor is passed that is already created with the arguments or a model’s config an error is thrown.
• stopping_criteria (StoppingCriteriaList, optional) — Custom stopping criteria that complement the default stopping criteria built from arguments and a model’s config. If a stopping criteria is passed that is already created with the arguments or a model’s config an error is thrown.
• forced_bos_token_id (int, optional) — 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, optional) — The id of the token to force as the last generated token when max_length is reached.
• remove_invalid_values (bool, optional) — 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.

Returns

torch.LongTensor of shape (batch_size * num_return_sequences, sequence_length)

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.

Implements RAG token decoding.

## TFRagModel

### class transformers.TFRagModel

< >

( *args **kwargs )

Parameters

• config (RagConfig) — Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.
• question_encoder (TFPreTrainedModel) — An encoder model compatible with the faiss index encapsulated by the retriever.
• generator (TFPreTrainedModel) — A seq2seq model used as the generator in the RAG architecture.
• retriever (RagRetriever) — A retriever class encapsulating a faiss index queried to obtain context documents for current inputs.

The TFRagModel forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

RAG is a sequence-to-sequence model which encapsulates two core components: a question encoder and a generator. During a forward pass, we encode the input with the question encoder and pass it to the retriever to extract relevant context documents. The documents are then prepended to the input. Such contextualized inputs is passed to the generator.

The question encoder can be any autoencoding model, preferably TFDPRQuestionEncoder, and the generator can be any seq2seq model, preferably TFBartForConditionalGeneration.

The model can be initialized with a RagRetriever for end-to-end generation or used in combination with the outputs of a retriever in multiple steps---see examples for more details. The model is compatible any autoencoding model as the question_encoder and any seq2seq model with language model head as the generator. It has been tested with TFDPRQuestionEncoder as the question_encoder and TFBartForConditionalGeneration as the generator.

This model inherits from TFPreTrainedModel. Check the superclass documentation for the generic methods the library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads etc.)

This model is also a Tensorflow tf.keras.Model subclass. Use it as a regular TF 2.0 Keras Model and refer to the TF 2.0 documentation for all matter related to general usage and behavior.

The model is in a developing state as it is now fully supports in eager-mode only, and may not be exported in SavedModel format.

#### call

< >

( input_ids = None attention_mask = None encoder_outputs = None decoder_input_ids = None decoder_attention_mask = None past_key_values = None doc_scores = None context_input_ids = None context_attention_mask = None use_cache = None output_attentions = None output_hidden_states = None output_retrieved = None n_docs = None return_dict = None training = False **kwargs ) transformers.models.rag.modeling_tf_rag.TFRetrievAugLMOutputor tuple(tf.Tensor)

Parameters

• input_ids (tf.Tensor of shape (batch_size, sequence_length)) — Indices of input sequence tokens in the vocabulary. RagConfig, used to initialize the model, specifies which generator to use, it also specifies a compatible generator tokenizer. Use that tokenizer class to obtain the indices.
• attention_mask (tf.Tensor of shape (batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

• 1 for tokens that are not masked,
• 0 for tokens that are masked.

• encoder_outputs (tuple(tuple(tf.Tensor), optional) — Tuple consists of (generator_enc_last_hidden_state, optional: generator_enc_hidden_states, optional: generator_enc_attentions). generator_enc_last_hidden_state of shape (batch_size, n_docs * sequence_length, hidden_size) is a sequence of hidden-states at the output of the last layer of the generator’s encoder.

Used by the (TFRagModel) model during decoding.

• decoder_input_ids (tf.Tensor of shape (batch_size, target_sequence_length), optional) — Provide for generation tasks. None by default, construct as per instructions for the generator model you’re using with your RAG instance.
• decoder_attention_mask (torch.BoolTensor of shape (batch_size, target_sequence_length), optional) — Default behavior: generate a tensor that ignores pad tokens in decoder_input_ids. Causal mask will also be used by default.
• past_key_values (tuple(tuple(tf.Tensor))) — Tuple consists of two elements: encoder_outputs of the RAG model (see encoder_outputs) and past_key_values of the underlying generator. Can be used to speed up decoding. past_key_values are used in the (RagTokenForGeneration) model during decoding.
• doc_scores (tf.Tensor of shape (batch_size, config.n_docs)) — Score between each retrieved document embeddings (see retrieved_doc_embeds) and question_encoder_last_hidden_state. If the model has is not initialized with a retriever doc_scores has to be provided to the forward pass. doc_scores can be computed via question_encoder_last_hidden_state and retrieved_doc_embeds, see examples for more information.
• context_input_ids (tf.Tensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Input IDs post-processed from the retrieved documents and the question encoder input_ids by the retriever.

If the model has is not initialized with a retriever context_input_ids has to be provided to the forward pass. context_input_ids are returned by __call__() context_attention_mask (tf.Tensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True): Attention mask post-processed from the retrieved documents and the question encoder input_ids by the retriever.

If the model has is not initialized with a retriever context_attention_mask has to be provided to the forward pass. context_attention_mask are returned by __call__()

• use_cache (bool, optional, defaults to True) — If set to True, past_key_values key value states are returned and can be used to speed up decoding (see past_key_values).
• output_attentions (bool, optional) — Whether or not to return the attentions tensors of all attention layers. See attentions under returned tensors for more detail.
• output_hidden_states (bool, optional) — Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for more detail.
• output_retrieved(bool, optional) — Whether or not to return the retrieved_doc_embeds, retrieved_doc_ids, context_input_ids and context_attention_mask. See returned tensors for more detail.
• return_dict (bool, optional) — Whether or not to return a TFRetrievAugLMOutput instead of a plain tuple.
• n_docs (int, optional, defaults to config.n_docs“) — Number of documents to retrieve and/or number of documents for which to generate an answer.

Returns

transformers.models.rag.modeling_tf_rag.TFRetrievAugLMOutputor tuple(tf.Tensor)

A transformers.models.rag.modeling_tf_rag.TFRetrievAugLMOutputor a tuple of tf.Tensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (RagConfig) and inputs.

• logits (tf.Tensor of shape (batch_size, sequence_length, config.vocab_size)) — Prediction scores of the language modeling head. The score is possibly marginalized over all documents for each vocabulary token.

• past_key_values (List[tf.Tensor], optional, returned when use_cache=True is passed or when config.use_cache=True) — List of tf.Tensor of length config.n_layers, with each tensor of shape (2, batch_size, num_heads, sequence_length, embed_size_per_head)).

Contains precomputed hidden-states (key and values in the attention blocks) of the decoder that can be used (see past_key_values input) to speed up sequential decoding.

• doc_scores (tf.Tensor of shape (batch_size, config.n_docs)) — Score between each retrieved document embeddings (see retrieved_doc_embeds) and question_encoder_last_hidden_state.

• retrieved_doc_embeds (tf.Tensor of shape (batch_size, config.n_docs, hidden_size), optional, returned when output_retrieved=True) — Embedded documents retrieved by the retriever. Is used with question_encoder_last_hidden_state to compute the doc_scores.

• retrieved_doc_ids (tf.Tensor of shape (batch_size, config.n_docs), optional, returned when output_retrieved=True) — The indexes of the embedded documents retrieved by the retriever.

• context_input_ids (tf.Tensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Input ids post-processed from the retrieved documents and the question encoder input_ids by the retriever.

• context_attention_mask (tf.Tensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Attention mask post-processed from the retrieved documents and the question encoder input_ids by the retriever.

• question_encoder_last_hidden_state (tf.Tensor of shape (batch_size, sequence_length, hidden_size), optional) — Sequence of hidden states at the output of the last layer of the question encoder pooled output of the model.

• question_enc_hidden_states (tuple(tf.Tensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of tf.Tensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the question encoder at the output of each layer plus the initial embedding outputs.

• question_enc_attentions (tuple(tf.Tensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of tf.Tensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the question encoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_enc_last_hidden_state (tf.Tensor of shape (batch_size, sequence_length, hidden_size), optional) — Sequence of hidden-states at the output of the last layer of the generator encoder of the model.

• generator_enc_hidden_states (tuple(tf.Tensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of tf.Tensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the generator encoder at the output of each layer plus the initial embedding outputs.

• generator_enc_attentions (tuple(tf.Tensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of tf.Tensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the generator encoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_dec_hidden_states (tuple(tf.Tensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of tf.Tensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the generator decoder at the output of each layer plus the initial embedding outputs.

• generator_dec_attentions (tuple(tf.Tensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of tf.Tensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the generator decoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

The TFRagModel forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example:

>>> from transformers import RagTokenizer, RagRetriever, TFRagModel
>>> import torch

>>> retriever = RagRetriever.from_pretrained(
... )
>>> # initialize with RagRetriever to do everything in one forward call
>>> model = TFRagModel.from_pretrained("facebook/rag-token-base", retriever=retriever, from_pt=True)

>>> input_dict = tokenizer.prepare_seq2seq_batch(
...     "How many people live in Paris?", "In Paris, there are 10 million people.", return_tensors="tf"
... )
>>> input_ids = input_dict["input_ids"]
>>> outputs = model(input_ids)

## TFRagSequenceForGeneration

### class transformers.TFRagSequenceForGeneration

< >

( *args **kwargs )

Parameters

• config (RagConfig) — Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.
• question_encoder (TFPreTrainedModel) — An encoder model compatible with the faiss index encapsulated by the retriever.
• generator (TFPreTrainedModel) — A seq2seq model used as the generator in the RAG architecture.
• retriever (RagRetriever) — A retriever class encapsulating a faiss index queried to obtain context documents for current inputs.

The TFRagSequenceForGeneration forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

A TF RAG-sequence model implementation. It performs RAG-sequence specific marginalization in the forward pass.

RAG is a sequence-to-sequence model which encapsulates two core components: a question encoder and a generator. During a forward pass, we encode the input with the question encoder and pass it to the retriever to extract relevant context documents. The documents are then prepended to the input. Such contextualized inputs is passed to the generator.

The question encoder can be any autoencoding model, preferably TFDPRQuestionEncoder, and the generator can be any seq2seq model, preferably TFBartForConditionalGeneration.

The model can be initialized with a RagRetriever for end-to-end generation or used in combination with the outputs of a retriever in multiple steps---see examples for more details. The model is compatible any autoencoding model as the question_encoder and any seq2seq model with language model head as the generator. It has been tested with TFDPRQuestionEncoder as the question_encoder and TFBartForConditionalGeneration as the generator.

This model inherits from TFPreTrainedModel. Check the superclass documentation for the generic methods the library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads etc.)

This model is also a Tensorflow tf.keras.Model subclass. Use it as a regular TF 2.0 Keras Model and refer to the TF 2.0 documentation for all matter related to general usage and behavior.

The model is in a developing state as it is now fully supports in eager-mode only, and may not be exported in SavedModel format.

#### call

< >

( input_ids = None attention_mask = None decoder_input_ids = None decoder_attention_mask = None encoder_outputs = None past_key_values = None doc_scores = None context_input_ids = None context_attention_mask = None use_cache = None output_attentions = None output_hidden_states = None output_retrieved = None n_docs = None exclude_bos_score = None labels = None reduce_loss = None return_dict = None training = False **kwargs ) transformers.models.rag.modeling_tf_rag.TFRetrievAugLMMarginOutputor tuple(tf.Tensor)

Parameters

• input_ids (tf.Tensor of shape (batch_size, sequence_length)) — Indices of input sequence tokens in the vocabulary. RagConfig, used to initialize the model, specifies which generator to use, it also specifies a compatible generator tokenizer. Use that tokenizer class to obtain the indices.
• attention_mask (tf.Tensor of shape (batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

• 1 for tokens that are not masked,
• 0 for tokens that are masked.

• encoder_outputs (tuple(tuple(tf.Tensor), optional) — Tuple consists of (generator_enc_last_hidden_state, optional: generator_enc_hidden_states, optional: generator_enc_attentions). generator_enc_last_hidden_state of shape (batch_size, n_docs * sequence_length, hidden_size) is a sequence of hidden-states at the output of the last layer of the generator’s encoder.

Used by the (TFRagModel) model during decoding.

• decoder_input_ids (tf.Tensor of shape (batch_size, target_sequence_length), optional) — Provide for generation tasks. None by default, construct as per instructions for the generator model you’re using with your RAG instance.
• decoder_attention_mask (torch.BoolTensor of shape (batch_size, target_sequence_length), optional) — Default behavior: generate a tensor that ignores pad tokens in decoder_input_ids. Causal mask will also be used by default.
• past_key_values (tuple(tuple(tf.Tensor))) — Tuple consists of two elements: encoder_outputs of the RAG model (see encoder_outputs) and past_key_values of the underlying generator. Can be used to speed up decoding. past_key_values are used in the (RagTokenForGeneration) model during decoding.
• doc_scores (tf.Tensor of shape (batch_size, config.n_docs)) — Score between each retrieved document embeddings (see retrieved_doc_embeds) and question_encoder_last_hidden_state. If the model has is not initialized with a retriever doc_scores has to be provided to the forward pass. doc_scores can be computed via question_encoder_last_hidden_state and retrieved_doc_embeds, see examples for more information.
• context_input_ids (tf.Tensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Input IDs post-processed from the retrieved documents and the question encoder input_ids by the retriever.

If the model has is not initialized with a retriever context_input_ids has to be provided to the forward pass. context_input_ids are returned by __call__() context_attention_mask (tf.Tensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True): Attention mask post-processed from the retrieved documents and the question encoder input_ids by the retriever.

If the model has is not initialized with a retriever context_attention_mask has to be provided to the forward pass. context_attention_mask are returned by __call__()

• use_cache (bool, optional, defaults to True) — If set to True, past_key_values key value states are returned and can be used to speed up decoding (see past_key_values).
• output_attentions (bool, optional) — Whether or not to return the attentions tensors of all attention layers. See attentions under returned tensors for more detail.
• output_hidden_states (bool, optional) — Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for more detail.
• output_retrieved(bool, optional) — Whether or not to return the retrieved_doc_embeds, retrieved_doc_ids, context_input_ids and context_attention_mask. See returned tensors for more detail.
• return_dict (bool, optional) — Whether or not to return a TFRetrievAugLMOutput instead of a plain tuple.
• n_docs (int, optional, defaults to config.n_docs“) — Number of documents to retrieve and/or number of documents for which to generate an answer.
• exclude_bos_score (bool, optional) — Only relevant if labels is passed. If True, the score of the BOS token is disregarded when computing the loss.
• labels (tf.Tensor or np.ndarray of shape (batch_size, sequence_length), optional) — Labels for computing the cross entropy classification loss according to Rag-Sequence model formulation See https://arxiv.org/pdf/2005.11401.pdf Section 2.1 for details about Rag-Sequence formulation. Indices should be in [0, ..., config.vocab_size - 1].
• reduce_loss (bool, optional) — Only relevant if labels is passed. If True, the NLL loss is reduced using the tf.Tensor.sum operation.
• kwargs (Dict[str, any], optional, defaults to {}) — Legacy dictionary, which is required so that model can use generate() function.

Returns

transformers.models.rag.modeling_tf_rag.TFRetrievAugLMMarginOutputor tuple(tf.Tensor)

A transformers.models.rag.modeling_tf_rag.TFRetrievAugLMMarginOutputor a tuple of tf.Tensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (RagConfig) and inputs.

• loss (tf.Tensor of shape (1,), optional, returned when labels is provided) — Language modeling loss.

• logits (tf.Tensor of shape (batch_size, sequence_length, config.vocab_size)) — Prediction scores of the language modeling head. The score is possibly marginalized over all documents for each vocabulary token.

• past_key_values (List[tf.Tensor], optional, returned when use_cache=True is passed or when config.use_cache=True) — List of tf.Tensor of length config.n_layers, with each tensor of shape (2, batch_size, num_heads, sequence_length, embed_size_per_head)).

Contains precomputed hidden-states (key and values in the attention blocks) of the decoder that can be used (see past_key_values input) to speed up sequential decoding.

• doc_scores (tf.Tensor of shape (batch_size, config.n_docs)) — Score between each retrieved document embeddings (see retrieved_doc_embeds) and question_encoder_last_hidden_state.

• retrieved_doc_embeds (tf.Tensor of shape (batch_size, config.n_docs, hidden_size), optional, returned when output_retrieved=True) — Embedded documents retrieved by the retriever. Is used with question_encoder_last_hidden_state to compute the doc_scores.

• retrieved_doc_ids (tf.Tensor (int32) of shape (batch_size, config.n_docs), optional, returned when output_retrieved=True) — The indexes of the embedded documents retrieved by the retriever.

• context_input_ids (tf.Tensor(int32) of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Input ids post-processed from the retrieved documents and the question encoder input_ids by the retriever.

• context_attention_mask (tf.Tensor (int32) of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Attention mask post-processed from the retrieved documents and the question encoder input_ids by the retriever.

• question_encoder_last_hidden_state (tf.Tensor of shape (batch_size, sequence_length, hidden_size), optional) — Sequence of hidden states at the output of the last layer of the question encoder pooled output of the model.

• question_enc_hidden_states (tuple(tf.Tensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of tf.Tensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the question encoder at the output of each layer plus the initial embedding outputs.

• question_enc_attentions (tuple(tf.Tensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of tf.Tensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the question encoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_enc_last_hidden_state (tf.Tensor of shape (batch_size, sequence_length, hidden_size), optional) — Sequence of hidden-states at the output of the last layer of the generator encoder of the model.

• generator_enc_hidden_states (tuple(tf.Tensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of tf.Tensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the generator encoder at the output of each layer plus the initial embedding outputs.

• generator_enc_attentions (tuple(tf.Tensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of tf.Tensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the generator encoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_dec_hidden_states (tuple(tf.Tensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of tf.Tensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the generator decoder at the output of each layer plus the initial embedding outputs.

• generator_dec_attentions (tuple(tf.Tensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of tf.Tensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the generator decoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

The TFRagSequenceForGeneration forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example:

>>> from transformers import RagTokenizer, RagRetriever, TFRagSequenceForGeneration

>>> retriever = RagRetriever.from_pretrained(
... )
>>> # initialize with RagRetriever to do everything in one forward call
>>> model = TFRagSequenceForGeneration.from_pretrained(
... )

>>> input_dict = tokenizer.prepare_seq2seq_batch(
...     "How many people live in Paris?", "In Paris, there are 10 million people.", return_tensors="tf"
... )
>>> outputs = model(input_dict, output_retrieved=True)

>>> # or use retriever separately
>>> # 1. Encode
>>> input_ids = input_dict["input_ids"]
>>> question_hidden_states = model.question_encoder(input_ids)[0]
>>> # 2. Retrieve
>>> docs_dict = retriever(input_ids.numpy(), question_hidden_states.numpy(), return_tensors="tf")
>>> doc_scores = tf.squeeze(
...     tf.matmul(
...         tf.expand_dims(question_hidden_states, axis=1), docs_dict["retrieved_doc_embeds"], transpose_b=True
...     ),
...     axis=1,
... )
>>> # 3. Forward to generator
>>> outputs = model(
...     inputs=None,
...     context_input_ids=docs_dict["context_input_ids"],
...     doc_scores=doc_scores,
...     decoder_input_ids=input_dict["labels"],
... )

>>> # or directly generate
>>> generated = model.generate(
...     context_input_ids=docs_dict["context_input_ids"],
...     doc_scores=doc_scores,
... )
>>> generated_string = tokenizer.batch_decode(generated, skip_special_tokens=True)

#### generate

< >

( input_ids: typing.Optional[tensorflow.python.framework.ops.Tensor] = None attention_mask: typing.Optional[tensorflow.python.framework.ops.Tensor] = None context_input_ids = None context_attention_mask = None doc_scores = None do_deduplication = None num_return_sequences = None num_beams = None n_docs = None **model_kwargs ) tf.Tensor of shape (batch_size * num_return_sequences, sequence_length)

Parameters

• input_ids (tf.Tensor of shape (batch_size, sequence_length), optional) — The sequence used as a prompt for the generation. If input_ids is not passed, then context_input_ids has to be provided.
• attention_mask (tf.Tensor of shape (batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]: - 1 for tokens that are not masked, - 0 for tokens that are masked. What are attention masks?
• context_input_ids (tf.Tensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Input IDs post-processed from the retrieved documents and the question encoder input_ids by the retriever.
• context_attention_mask (tf.Tensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Attention mask post-processed from the retrieved documents and the question encoder input_ids by the retriever. If the model has is not initialized with a retriever or input_ids is not given, context_input_ids and context_attention_mask have to be provided to the forward pass. They are returned by __call__()
• doc_scores (tf.Tensor of shape (batch_size, config.n_docs)) — Score between each retrieved document embeddings (see retrieved_doc_embeds) and question_encoder_last_hidden_state. If the model has is not initialized with a retriever or input_ids is not given, doc_scores has to be provided to the forward pass. doc_scores are returned by __call__()
• do_deduplication (bool, optional) — Whether or not to deduplicate the generations from different context documents for a given input. Has to be set to False if used while training with distributed backend.
• num_return_sequences(int, optional, defaults to 1) — The number of independently computed returned sequences for each element in the batch. Note that this is not the value we pass to the generator’s [generate()](/docs/transformers/v4.17.0/en/main_classes/model#transformers.generation_utils.GenerationMixin.generate) function, where we set num_return_sequences to num_beams.
• num_beams (int, optional, defaults to 1) — Number of beams for beam search. 1 means no beam search.
• n_docs (int, optional, defaults to config.n_docs) — Number of documents to retrieve and/or number of documents for which to generate an answer. kwargs — Additional kwargs will be passed to generate()

Returns

tf.Tensor of shape (batch_size * num_return_sequences, sequence_length)

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.

Implements RAG sequence “thorough” decoding. Read the generate() documentation for more information on how to set other generate input parameters

## TFRagTokenForGeneration

### class transformers.TFRagTokenForGeneration

< >

( *args **kwargs )

Parameters

• config (RagConfig) — Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.
• question_encoder (TFPreTrainedModel) — An encoder model compatible with the faiss index encapsulated by the retriever.
• generator (TFPreTrainedModel) — A seq2seq model used as the generator in the RAG architecture.
• retriever (RagRetriever) — A retriever class encapsulating a faiss index queried to obtain context documents for current inputs.

The TFRagTokenForGeneration forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

A TF RAG-token model implementation. It performs RAG-token specific marginalization in the forward pass.

RAG is a sequence-to-sequence model which encapsulates two core components: a question encoder and a generator. During a forward pass, we encode the input with the question encoder and pass it to the retriever to extract relevant context documents. The documents are then prepended to the input. Such contextualized inputs is passed to the generator.

The question encoder can be any autoencoding model, preferably TFDPRQuestionEncoder, and the generator can be any seq2seq model, preferably TFBartForConditionalGeneration.

The model can be initialized with a RagRetriever for end-to-end generation or used in combination with the outputs of a retriever in multiple steps---see examples for more details. The model is compatible any autoencoding model as the question_encoder and any seq2seq model with language model head as the generator. It has been tested with TFDPRQuestionEncoder as the question_encoder and TFBartForConditionalGeneration as the generator.

This model inherits from TFPreTrainedModel. Check the superclass documentation for the generic methods the library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads etc.)

This model is also a Tensorflow tf.keras.Model subclass. Use it as a regular TF 2.0 Keras Model and refer to the TF 2.0 documentation for all matter related to general usage and behavior.

The model is in a developing state as it is now fully supports in eager-mode only, and may not be exported in SavedModel format.

#### call

< >

( input_ids = None attention_mask = None decoder_input_ids = None decoder_attention_mask = None encoder_outputs = None past_key_values = None doc_scores = None context_input_ids = None context_attention_mask = None use_cache = None output_attentions = None output_hidden_states = None output_retrieved = None n_docs = None do_marginalize = None labels = None reduce_loss = None return_dict = None training = False **kwargs ) transformers.models.rag.modeling_tf_rag.TFRetrievAugLMMarginOutputor tuple(tf.Tensor)

Parameters

• input_ids (tf.Tensor of shape (batch_size, sequence_length)) — Indices of input sequence tokens in the vocabulary. RagConfig, used to initialize the model, specifies which generator to use, it also specifies a compatible generator tokenizer. Use that tokenizer class to obtain the indices.
• attention_mask (tf.Tensor of shape (batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

• 1 for tokens that are not masked,
• 0 for tokens that are masked.

• encoder_outputs (tuple(tuple(tf.Tensor), optional) — Tuple consists of (generator_enc_last_hidden_state, optional: generator_enc_hidden_states, optional: generator_enc_attentions). generator_enc_last_hidden_state of shape (batch_size, n_docs * sequence_length, hidden_size) is a sequence of hidden-states at the output of the last layer of the generator’s encoder.

Used by the (TFRagModel) model during decoding.

• decoder_input_ids (tf.Tensor of shape (batch_size, target_sequence_length), optional) — Provide for generation tasks. None by default, construct as per instructions for the generator model you’re using with your RAG instance.
• decoder_attention_mask (torch.BoolTensor of shape (batch_size, target_sequence_length), optional) — Default behavior: generate a tensor that ignores pad tokens in decoder_input_ids. Causal mask will also be used by default.
• past_key_values (tuple(tuple(tf.Tensor))) — Tuple consists of two elements: encoder_outputs of the RAG model (see encoder_outputs) and past_key_values of the underlying generator. Can be used to speed up decoding. past_key_values are used in the (RagTokenForGeneration) model during decoding.
• doc_scores (tf.Tensor of shape (batch_size, config.n_docs)) — Score between each retrieved document embeddings (see retrieved_doc_embeds) and question_encoder_last_hidden_state. If the model has is not initialized with a retriever doc_scores has to be provided to the forward pass. doc_scores can be computed via question_encoder_last_hidden_state and retrieved_doc_embeds, see examples for more information.
• context_input_ids (tf.Tensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Input IDs post-processed from the retrieved documents and the question encoder input_ids by the retriever.

If the model has is not initialized with a retriever context_input_ids has to be provided to the forward pass. context_input_ids are returned by __call__() context_attention_mask (tf.Tensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True): Attention mask post-processed from the retrieved documents and the question encoder input_ids by the retriever.

If the model has is not initialized with a retriever context_attention_mask has to be provided to the forward pass. context_attention_mask are returned by __call__()

• use_cache (bool, optional, defaults to True) — If set to True, past_key_values key value states are returned and can be used to speed up decoding (see past_key_values).
• output_attentions (bool, optional) — Whether or not to return the attentions tensors of all attention layers. See attentions under returned tensors for more detail.
• output_hidden_states (bool, optional) — Whether or not to return the hidden states of all layers. See hidden_states under returned tensors for more detail.
• output_retrieved(bool, optional) — Whether or not to return the retrieved_doc_embeds, retrieved_doc_ids, context_input_ids and context_attention_mask. See returned tensors for more detail.
• return_dict (bool, optional) — Whether or not to return a TFRetrievAugLMOutput instead of a plain tuple.
• n_docs (int, optional, defaults to config.n_docs“) — Number of documents to retrieve and/or number of documents for which to generate an answer.
• do_marginalize (bool, optional) — If True, the logits are marginalized over all documents by making use of torch.nn.functional.log_softmax.
• labels (tf.Tensor or np.ndarray of shape (batch_size, sequence_length), optional) — Labels for computing the cross entropy classification loss according to Rag-Token model formulation See https://arxiv.org/pdf/2005.11401.pdf Section 2.1 for details about Rag-Token formulation. Indices should be in [0, ..., config.vocab_size - 1].
• reduce_loss (bool, optional) — Only relevant if labels is passed. If True, the NLL loss is reduced using the tf.Tensor.sum operation.
• kwargs (Dict[str, any], optional, defaults to {}) — Legacy dictionary, which is required so that model can use generate() function.

Returns

transformers.models.rag.modeling_tf_rag.TFRetrievAugLMMarginOutputor tuple(tf.Tensor)

A transformers.models.rag.modeling_tf_rag.TFRetrievAugLMMarginOutputor a tuple of tf.Tensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (RagConfig) and inputs.

• loss (tf.Tensor of shape (1,), optional, returned when labels is provided) — Language modeling loss.

• logits (tf.Tensor of shape (batch_size, sequence_length, config.vocab_size)) — Prediction scores of the language modeling head. The score is possibly marginalized over all documents for each vocabulary token.

• past_key_values (List[tf.Tensor], optional, returned when use_cache=True is passed or when config.use_cache=True) — List of tf.Tensor of length config.n_layers, with each tensor of shape (2, batch_size, num_heads, sequence_length, embed_size_per_head)).

Contains precomputed hidden-states (key and values in the attention blocks) of the decoder that can be used (see past_key_values input) to speed up sequential decoding.

• doc_scores (tf.Tensor of shape (batch_size, config.n_docs)) — Score between each retrieved document embeddings (see retrieved_doc_embeds) and question_encoder_last_hidden_state.

• retrieved_doc_embeds (tf.Tensor of shape (batch_size, config.n_docs, hidden_size), optional, returned when output_retrieved=True) — Embedded documents retrieved by the retriever. Is used with question_encoder_last_hidden_state to compute the doc_scores.

• retrieved_doc_ids (tf.Tensor (int32) of shape (batch_size, config.n_docs), optional, returned when output_retrieved=True) — The indexes of the embedded documents retrieved by the retriever.

• context_input_ids (tf.Tensor(int32) of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Input ids post-processed from the retrieved documents and the question encoder input_ids by the retriever.

• context_attention_mask (tf.Tensor (int32) of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Attention mask post-processed from the retrieved documents and the question encoder input_ids by the retriever.

• question_encoder_last_hidden_state (tf.Tensor of shape (batch_size, sequence_length, hidden_size), optional) — Sequence of hidden states at the output of the last layer of the question encoder pooled output of the model.

• question_enc_hidden_states (tuple(tf.Tensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of tf.Tensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the question encoder at the output of each layer plus the initial embedding outputs.

• question_enc_attentions (tuple(tf.Tensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of tf.Tensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the question encoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_enc_last_hidden_state (tf.Tensor of shape (batch_size, sequence_length, hidden_size), optional) — Sequence of hidden-states at the output of the last layer of the generator encoder of the model.

• generator_enc_hidden_states (tuple(tf.Tensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of tf.Tensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the generator encoder at the output of each layer plus the initial embedding outputs.

• generator_enc_attentions (tuple(tf.Tensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of tf.Tensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the generator encoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

• generator_dec_hidden_states (tuple(tf.Tensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of tf.Tensor (one for the output of the embeddings and one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

Hidden states of the generator decoder at the output of each layer plus the initial embedding outputs.

• generator_dec_attentions (tuple(tf.Tensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of tf.Tensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

Attentions weights of the generator decoder, after the attention softmax, used to compute the weighted average in the self-attention heads.

The TFRagTokenForGeneration forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example:

>>> import tensorflow as tf
>>> from transformers import RagTokenizer, RagRetriever, TFRagTokenForGeneration

>>> retriever = RagRetriever.from_pretrained(
... )
>>> # initialize with RagRetriever to do everything in one forward call
>>> model = TFRagTokenForGeneration.from_pretrained("facebook/rag-token-nq", retriever=retriever, from_pt=True)

>>> input_dict = tokenizer.prepare_seq2seq_batch(
...     "How many people live in Paris?", "In Paris, there are 10 million people.", return_tensors="tf"
... )
>>> outputs = model(input_dict, output_retrieved=True)

>>> # or use retriever separately
>>> # 1. Encode
>>> input_ids = input_dict["input_ids"]
>>> question_hidden_states = model.question_encoder(input_ids)[0]
>>> # 2. Retrieve
>>> docs_dict = retriever(input_ids.numpy(), question_hidden_states.numpy(), return_tensors="tf")
>>> doc_scores = tf.squeeze(
...     tf.matmul(
...         tf.expand_dims(question_hidden_states, axis=1), docs_dict["retrieved_doc_embeds"], transpose_b=True
...     ),
...     axis=1,
... )
>>> # 3. Forward to generator
>>> outputs = model(
...     inputs=None,
...     context_input_ids=docs_dict["context_input_ids"],
...     doc_scores=doc_scores,
...     decoder_input_ids=input_dict["labels"],
... )

>>> # or directly generate
>>> generated = model.generate(
...     context_input_ids=docs_dict["context_input_ids"],
...     doc_scores=doc_scores,
... )
>>> generated_string = tokenizer.batch_decode(generated, skip_special_tokens=True)

#### generate

< >

( input_ids: typing.Optional[tensorflow.python.framework.ops.Tensor] = None attention_mask: typing.Optional[tensorflow.python.framework.ops.Tensor] = None context_input_ids = None context_attention_mask = None doc_scores = None max_length = None min_length = None early_stopping = None use_cache = None num_beams = None bos_token_id = None pad_token_id = None eos_token_id = None length_penalty = None no_repeat_ngram_size = None bad_words_ids = None num_return_sequences = None decoder_start_token_id = None n_docs = None output_scores = None output_attentions = None output_hidden_states = None return_dict_in_generate = None **model_kwargs ) tf.Tensor of shape (batch_size * num_return_sequences, sequence_length)

Parameters

• input_ids (tf.Tensor of shape (batch_size, sequence_length), optional) — The sequence used as a prompt for the generation. If input_ids is not passed, then context_input_ids has to be provided.
• attention_mask (tf.Tensor of shape (batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

• 1 for tokens that are not masked,
• 0 for tokens that are masked.

• context_input_ids (tf.Tensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Input IDs post-processed from the retrieved documents and the question encoder input_ids by the retriever.

If the model has is not initialized with a retriever, context_input_ids has to be provided to the forward pass. context_input_ids are returned by __call__()

• context_attention_mask (tf.Tensor of shape (batch_size * config.n_docs, config.max_combined_length), optional, returned when output_retrieved=True) — Attention mask post-processed from the retrieved documents and the question encoder input_ids by the retriever.

If the model has is not initialized with a retriever, context_input_ids has to be provided to the forward pass. context_input_ids are returned by __call__()

• doc_scores (tf.Tensor of shape (batch_size, config.n_docs)) — Score between each retrieved document embeddings (see retrieved_doc_embeds) and question_encoder_last_hidden_state.

If the model has is not initialized with a retriever, context_input_ids has to be provided to the forward pass. context_input_ids are returned by __call__()

• max_length (int, optional, defaults to 20) — The maximum length of the sequence to be generated.
• min_length (int, optional, defaults to 10) — The minimum length of the sequence to be generated.
• early_stopping (bool, optional, defaults to False) — Whether or not to stop the beam search when at least num_beams sentences are finished per batch or not. 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.
• 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 (int, optional) — The id of the end-of-sequence token.
• length_penalty (float, optional, defaults to 1.0) — Exponential penalty to the length. 1.0 means no penalty.

Set to values < 1.0 in order to encourage the model to generate shorter sequences, to a value > 1.0 in order to encourage the model to produce longer 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[int], optional) — List of token ids that are not allowed to be generated. In order to get the tokens of the words that should not appear in the generated text, use tokenizer.encode(bad_word, add_prefix_space=True).
• num_beams (int, optional, defaults to 1) — Number of beams for beam search. 1 means no beam search.
• num_return_sequences(int, optional, defaults to 1) — The number of independently computed returned sequences for each element in the batch. Note that this is not the value we pass to the generator’s [generate()](/docs/transformers/v4.17.0/en/main_classes/model#transformers.generation_utils.GenerationMixin.generate) function, where we set num_return_sequencestonum_beams. decoder_start_token_id (int, optional): If an encoder-decoder model starts decoding with a different token than bos, the id of that token.
• n_docs (int, optional, defaults to config.n_docs) — Number of documents to retrieve and/or number of documents for which to generate an answer.
• 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.
• return_dict_in_generate (bool, optional, defaults to False) — Whether or not to return a ModelOutput instead of a plain tuple. model_specific_kwargs — Additional model specific kwargs will be forwarded to the forward function of the model.

Returns

tf.Tensor of shape (batch_size * num_return_sequences, sequence_length)

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.

Implements TFRAG token decoding.