# ESM

## ESM

### Overview

This page provides code and pre-trained weights for Transformer protein language models from Meta AI’s Fundamental AI Research Team, providing the state-of-the-art ESMFold and ESM-2, and the previously released ESM-1b and ESM-1v. Transformer protein language models were introduced in the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. The first version of this paper was [preprinted in 2019](https://www.biorxiv.org/content/10.1101/622803v1?versioned=true).

ESM-2 outperforms all tested single-sequence protein language models across a range of structure prediction tasks, and enables atomic resolution structure prediction. It was released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido and Alexander Rives.

Also introduced in this paper was ESMFold. It uses an ESM-2 stem with a head that can predict folded protein structures with state-of-the-art accuracy. Unlike [AlphaFold2](https://www.nature.com/articles/s41586-021-03819-2), it relies on the token embeddings from the large pre-trained protein language model stem and does not perform a multiple sequence alignment (MSA) step at inference time, which means that ESMFold checkpoints are fully “standalone” - they do not require a database of known protein sequences and structures with associated external query tools to make predictions, and are much faster as a result.

The abstract from “Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences” is

*In the field of artificial intelligence, a combination of scale in data and model capacity enabled by unsupervised learning has led to major advances in representation learning and statistical generation. In the life sciences, the anticipated growth of sequencing promises unprecedented data on natural sequence diversity. Protein language modeling at the scale of evolution is a logical step toward predictive and generative artificial intelligence for biology. To this end, we use unsupervised learning to train a deep contextual language model on 86 billion amino acids across 250 million protein sequences spanning evolutionary diversity. The resulting model contains information about biological properties in its representations. The representations are learned from sequence data alone. The learned representation space has a multiscale organization reflecting structure from the level of biochemical properties of amino acids to remote homology of proteins. Information about secondary and tertiary structure is encoded in the representations and can be identified by linear projections. Representation learning produces features that generalize across a range of applications, enabling state-of-the-art supervised prediction of mutational effect and secondary structure and improving state-of-the-art features for long-range contact prediction.*

The abstract from “Language models of protein sequences at the scale of evolution enable accurate structure prediction” is

*Large language models have recently been shown to develop emergent capabilities with scale, going beyond simple pattern matching to perform higher level reasoning and generate lifelike images and text. While language models trained on protein sequences have been studied at a smaller scale, little is known about what they learn about biology as they are scaled up. In this work we train models up to 15 billion parameters, the largest language models of proteins to be evaluated to date. We find that as models are scaled they learn information enabling the prediction of the three-dimensional structure of a protein at the resolution of individual atoms. We present ESMFold for high accuracy end-to-end atomic level structure prediction directly from the individual sequence of a protein. ESMFold has similar accuracy to AlphaFold2 and RoseTTAFold for sequences with low perplexity that are well understood by the language model. ESMFold inference is an order of magnitude faster than AlphaFold2, enabling exploration of the structural space of metagenomic proteins in practical timescales.*

Tips:

* ESM models are trained with a masked language modeling (MLM) objective.

The original code can be found [here](https://github.com/facebookresearch/esm) and was was developed by the Fundamental AI Research team at Meta AI. ESM-1b, ESM-1v and ESM-2 were contributed to boincai by [jasonliu](https://huggingface.co/jasonliu) and [Matt](https://huggingface.co/Rocketknight1).

ESMFold was contributed to boincai by [Matt](https://huggingface.co/Rocketknight1) and [Sylvain](https://huggingface.co/sgugger), with a big thank you to Nikita Smetanin, Roshan Rao and Tom Sercu for their help throughout the process!

The BOINCAI port of ESMFold uses portions of the [openfold](https://github.com/aqlaboratory/openfold) library. The `openfold` library is licensed under the Apache License 2.0.

### Documentation resources

* [Text classification task guide](https://huggingface.co/docs/transformers/tasks/sequence_classification)
* [Token classification task guide](https://huggingface.co/docs/transformers/tasks/token_classification)
* [Masked language modeling task guide](https://huggingface.co/docs/transformers/tasks/masked_language_modeling)

### EsmConfig

#### class transformers.EsmConfig

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/configuration_esm.py#L33)

( vocab\_size = Nonemask\_token\_id = Nonepad\_token\_id = Nonehidden\_size = 768num\_hidden\_layers = 12num\_attention\_heads = 12intermediate\_size = 3072hidden\_dropout\_prob = 0.1attention\_probs\_dropout\_prob = 0.1max\_position\_embeddings = 1026initializer\_range = 0.02layer\_norm\_eps = 1e-12position\_embedding\_type = 'absolute'use\_cache = Trueemb\_layer\_norm\_before = Nonetoken\_dropout = Falseis\_folding\_model = Falseesmfold\_config = Nonevocab\_list = None\*\*kwargs )

Parameters

* **vocab\_size** (`int`, *optional*) — Vocabulary size of the ESM model. Defines the number of different tokens that can be represented by the `inputs_ids` passed when calling `ESMModel`.
* **mask\_token\_id** (`int`, *optional*) — The index of the mask token in the vocabulary. This must be included in the config because of the “mask-dropout” scaling trick, which will scale the inputs depending on the number of masked tokens.
* **pad\_token\_id** (`int`, *optional*) — The index of the padding token in the vocabulary. This must be included in the config because certain parts of the ESM code use this instead of the attention mask.
* **hidden\_size** (`int`, *optional*, defaults to 768) — Dimensionality of the encoder layers and the pooler layer.
* **num\_hidden\_layers** (`int`, *optional*, defaults to 12) — Number of hidden layers in the Transformer encoder.
* **num\_attention\_heads** (`int`, *optional*, defaults to 12) — Number of attention heads for each attention layer in the Transformer encoder.
* **intermediate\_size** (`int`, *optional*, defaults to 3072) — Dimensionality of the “intermediate” (often named feed-forward) layer in the Transformer encoder.
* **hidden\_dropout\_prob** (`float`, *optional*, defaults to 0.1) — The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
* **attention\_probs\_dropout\_prob** (`float`, *optional*, defaults to 0.1) — The dropout ratio for the attention probabilities.
* **max\_position\_embeddings** (`int`, *optional*, defaults to 1026) — The maximum sequence length that this model might ever be used with. Typically set this to something large just in case (e.g., 512 or 1024 or 2048).
* **initializer\_range** (`float`, *optional*, defaults to 0.02) — The standard deviation of the truncated\_normal\_initializer for initializing all weight matrices.
* **layer\_norm\_eps** (`float`, *optional*, defaults to 1e-12) — The epsilon used by the layer normalization layers.
* **position\_embedding\_type** (`str`, *optional*, defaults to `"absolute"`) — Type of position embedding. Choose one of `"absolute"`, `"relative_key"`, `"relative_key_query", "rotary"`. For positional embeddings use `"absolute"`. For more information on `"relative_key"`, please refer to [Self-Attention with Relative Position Representations (Shaw et al.)](https://arxiv.org/abs/1803.02155). For more information on `"relative_key_query"`, please refer to *Method 4* in [Improve Transformer Models with Better Relative Position Embeddings (Huang et al.)](https://arxiv.org/abs/2009.13658).
* **is\_decoder** (`bool`, *optional*, defaults to `False`) — Whether the model is used as a decoder or not. If `False`, the model is used as an encoder.
* **use\_cache** (`bool`, *optional*, defaults to `True`) — Whether or not the model should return the last key/values attentions (not used by all models). Only relevant if `config.is_decoder=True`.
* **emb\_layer\_norm\_before** (`bool`, *optional*) — Whether to apply layer normalization after embeddings but before the main stem of the network.
* **token\_dropout** (`bool`, defaults to `False`) — When this is enabled, masked tokens are treated as if they had been dropped out by input dropout.

This is the configuration class to store the configuration of a `ESMModel`. It is used to instantiate a ESM model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of the ESM [facebook/esm-1b](https://huggingface.co/facebook/esm-1b) architecture.

Configuration objects inherit from [PretrainedConfig](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/configuration#transformers.PretrainedConfig) and can be used to control the model outputs. Read the documentation from [PretrainedConfig](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/configuration#transformers.PretrainedConfig) for more information.

Examples:

Copied

```
>>> from transformers import EsmModel, EsmConfig

>>> # Initializing a ESM facebook/esm-1b style configuration >>> configuration = EsmConfig()

>>> # Initializing a model from the configuration >>> model = ESMModel(configuration)

>>> # Accessing the model configuration >>> configuration = model.config
```

**to\_dict**

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/configuration_esm.py#L160)

( ) → `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()](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/configuration#transformers.PretrainedConfig.to_dict).

### EsmTokenizer

#### class transformers.EsmTokenizer

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/tokenization_esm.py#L47)

( vocab\_fileunk\_token = '\<unk>'cls\_token = '\<cls>'pad\_token = '\<pad>'mask\_token = '\<mask>'eos\_token = '\<eos>'\*\*kwargs )

Constructs an ESM tokenizer.

**build\_inputs\_with\_special\_tokens**

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/tokenization_esm.py#L106)

( token\_ids\_0: typing.List\[int]token\_ids\_1: typing.Optional\[typing.List\[int]] = None )

**get\_special\_tokens\_mask**

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/tokenization_esm.py#L120)

( token\_ids\_0: typing.Listtoken\_ids\_1: typing.Optional\[typing.List] = Nonealready\_has\_special\_tokens: bool = False ) → A list of integers in the range \[0, 1]

Parameters

* **token\_ids\_0** (`List[int]`) — List of ids of the first sequence.
* **token\_ids\_1** (`List[int]`, *optional*) — List of ids of the second sequence.
* **already\_has\_special\_tokens** (`bool`, *optional*, defaults to `False`) — Whether or not the token list is already formatted with special tokens for the model.

Returns

A list of integers in the range \[0, 1]

1 for a special token, 0 for a sequence token.

Retrieves sequence ids from a token list that has no special tokens added. This method is called when adding special tokens using the tokenizer `prepare_for_model` or `encode_plus` methods.

**create\_token\_type\_ids\_from\_sequences**

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/tokenization_utils_base.py#L3289)

( token\_ids\_0: typing.List\[int]token\_ids\_1: typing.Optional\[typing.List\[int]] = None ) → `List[int]`

Parameters

* **token\_ids\_0** (`List[int]`) — The first tokenized sequence.
* **token\_ids\_1** (`List[int]`, *optional*) — The second tokenized sequence.

Returns

`List[int]`

The token type ids.

Create the token type IDs corresponding to the sequences passed. [What are token type IDs?](https://huggingface.co/docs/transformers/glossary#token-type-ids)

Should be overridden in a subclass if the model has a special way of building those.

**save\_vocabulary**

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/tokenization_esm.py#L151)

( save\_directoryfilename\_prefix )

### EsmModel

#### class transformers.EsmModel

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/modeling_esm.py#L780)

( configadd\_pooling\_layer = True )

Parameters

* **config** ([EsmConfig](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmConfig)) — 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()](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights.

The bare ESM Model transformer outputting raw hidden-states without any specific head on top.

This model inherits from [PreTrainedModel](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/model#transformers.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](https://pytorch.org/docs/stable/nn.html#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.

The model can behave as an encoder (with only self-attention) as well as a decoder, in which case a layer of cross-attention is added between the self-attention layers, following the architecture described in [Attention is all you need](https://arxiv.org/abs/1706.03762) by Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser and Illia Polosukhin.

To behave as an decoder the model needs to be initialized with the `is_decoder` argument of the configuration set to `True`. To be used in a Seq2Seq model, the model needs to initialized with both `is_decoder` argument and `add_cross_attention` set to `True`; an `encoder_hidden_states` is then expected as an input to the forward pass.

**forward**

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/modeling_esm.py#L823)

( input\_ids: typing.Optional\[torch.Tensor] = Noneattention\_mask: typing.Optional\[torch.Tensor] = Noneposition\_ids: typing.Optional\[torch.Tensor] = Nonehead\_mask: typing.Optional\[torch.Tensor] = Noneinputs\_embeds: typing.Optional\[torch.Tensor] = Noneencoder\_hidden\_states: typing.Optional\[torch.Tensor] = Noneencoder\_attention\_mask: typing.Optional\[torch.Tensor] = Nonepast\_key\_values: typing.Optional\[typing.List\[torch.FloatTensor]] = Noneuse\_cache: typing.Optional\[bool] = Noneoutput\_attentions: typing.Optional\[bool] = Noneoutput\_hidden\_states: typing.Optional\[bool] = Nonereturn\_dict: typing.Optional\[bool] = None ) → [transformers.modeling\_outputs.BaseModelOutputWithPoolingAndCrossAttentions](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPoolingAndCrossAttentions) or `tuple(torch.FloatTensor)`

Parameters

* **input\_ids** (`torch.LongTensor` of shape `((batch_size, sequence_length))`) — Indices of input sequence tokens in the vocabulary.

  Indices can be obtained using [AutoTokenizer](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/tokenizer#transformers.PreTrainedTokenizerFast.encode) and [PreTrainedTokenizer.**call**()](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/vits#transformers.VitsTokenizer.__call__) for details.

  [What are input IDs?](https://huggingface.co/docs/transformers/glossary#input-ids)
* **attention\_mask** (`torch.FloatTensor` 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?](https://huggingface.co/docs/transformers/glossary#attention-mask)
* **position\_ids** (`torch.LongTensor` of shape `((batch_size, sequence_length))`, *optional*) — Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0, config.max_position_embeddings - 1]`.

  [What are position IDs?](https://huggingface.co/docs/transformers/glossary#position-ids)
* **head\_mask** (`torch.FloatTensor` of shape `(num_heads,)` or `(num_layers, num_heads)`, *optional*) — Mask to nullify selected heads of the self-attention modules. Mask values selected in `[0, 1]`:
  * 1 indicates the head is **not masked**,
  * 0 indicates the head is **masked**.
* **inputs\_embeds** (`torch.FloatTensor` of shape `((batch_size, sequence_length), hidden_size)`, *optional*) — Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert `input_ids` indices into associated vectors than the model’s internal embedding lookup matrix.
* **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.
* **return\_dict** (`bool`, *optional*) — Whether or not to return a [ModelOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.utils.ModelOutput) instead of a plain tuple.
* **encoder\_hidden\_states** (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*) — Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention if the model is configured as a decoder.
* **encoder\_attention\_mask** (`torch.FloatTensor` of shape `(batch_size, sequence_length)`, *optional*) — Mask to avoid performing attention on the padding token indices of the encoder input. This mask is used in the cross-attention if the model is configured as a decoder. Mask values selected in `[0, 1]`:
  * 1 for tokens that are **not masked**,
  * 0 for tokens that are **masked**.
* **past\_key\_values** (`tuple(tuple(torch.FloatTensor))` of length `config.n_layers` with each tuple having 4 tensors of shape `(batch_size, num_heads, sequence_length - 1, embed_size_per_head)`) — Contains precomputed key and value hidden states of the attention blocks. Can be used to speed up decoding.

  If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids` (those that don’t have their past key value states given to this model) of shape `(batch_size, 1)` instead of all `decoder_input_ids` of shape `(batch_size, sequence_length)`.
* **use\_cache** (`bool`, *optional*) — If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see `past_key_values`).

Returns

[transformers.modeling\_outputs.BaseModelOutputWithPoolingAndCrossAttentions](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPoolingAndCrossAttentions) or `tuple(torch.FloatTensor)`

A [transformers.modeling\_outputs.BaseModelOutputWithPoolingAndCrossAttentions](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPoolingAndCrossAttentions) 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 ([EsmConfig](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmConfig)) and inputs.

* **last\_hidden\_state** (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`) — Sequence of hidden-states at the output of the last layer of the model.
* **pooler\_output** (`torch.FloatTensor` of shape `(batch_size, hidden_size)`) — Last layer hidden-state of the first token of the sequence (classification token) after further processing through the layers used for the auxiliary pretraining task. E.g. for BERT-family of models, this returns the classification token after processing through a linear layer and a tanh activation function. The linear layer weights are trained from the next sentence prediction (classification) objective during pretraining.
* **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, if the model has an embedding layer, + one for the output of each layer) of shape `(batch_size, sequence_length, hidden_size)`.

  Hidden-states of the model at the output of each layer plus the optional initial embedding outputs.
* **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 after the attention softmax, used to compute the weighted average in the self-attention heads.
* **cross\_attentions** (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` and `config.add_cross_attention=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 decoder’s cross-attention layer, after the attention softmax, used to compute the weighted average in the cross-attention heads.
* **past\_key\_values** (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`) — Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2 tensors of shape `(batch_size, num_heads, sequence_length, embed_size_per_head)`) and optionally if `config.is_encoder_decoder=True` 2 additional tensors of shape `(batch_size, num_heads, encoder_sequence_length, embed_size_per_head)`.

  Contains pre-computed hidden-states (key and values in the self-attention blocks and optionally if `config.is_encoder_decoder=True` in the cross-attention blocks) that can be used (see `past_key_values` input) to speed up sequential decoding.

The [EsmModel](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmModel) 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:

Copied

```
>>> from transformers import AutoTokenizer, EsmModel
>>> import torch

>>> tokenizer = AutoTokenizer.from_pretrained("facebook/esm2_t6_8M_UR50D")
>>> model = EsmModel.from_pretrained("facebook/esm2_t6_8M_UR50D")

>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt")
>>> outputs = model(**inputs)

>>> last_hidden_states = outputs.last_hidden_state
```

### EsmForMaskedLM

#### class transformers.EsmForMaskedLM

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/modeling_esm.py#L963)

( config )

Parameters

* **config** ([EsmConfig](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmConfig)) — 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()](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights.

ESM Model with a `language modeling` head on top.

This model inherits from [PreTrainedModel](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/model#transformers.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](https://pytorch.org/docs/stable/nn.html#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**

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/modeling_esm.py#L986)

( input\_ids: typing.Optional\[torch.LongTensor] = Noneattention\_mask: typing.Optional\[torch.Tensor] = Noneposition\_ids: typing.Optional\[torch.LongTensor] = Nonehead\_mask: typing.Optional\[torch.Tensor] = Noneinputs\_embeds: typing.Optional\[torch.FloatTensor] = Noneencoder\_hidden\_states: typing.Optional\[torch.FloatTensor] = Noneencoder\_attention\_mask: typing.Optional\[torch.Tensor] = Nonelabels: typing.Optional\[torch.LongTensor] = Noneoutput\_attentions: typing.Optional\[bool] = Noneoutput\_hidden\_states: typing.Optional\[bool] = Nonereturn\_dict: typing.Optional\[bool] = None ) → [transformers.modeling\_outputs.MaskedLMOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_outputs.MaskedLMOutput) or `tuple(torch.FloatTensor)`

Parameters

* **input\_ids** (`torch.LongTensor` of shape `(batch_size, sequence_length)`) — Indices of input sequence tokens in the vocabulary.

  Indices can be obtained using [AutoTokenizer](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/tokenizer#transformers.PreTrainedTokenizerFast.encode) and [PreTrainedTokenizer.**call**()](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/vits#transformers.VitsTokenizer.__call__) for details.

  [What are input IDs?](https://huggingface.co/docs/transformers/glossary#input-ids)
* **attention\_mask** (`torch.FloatTensor` 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?](https://huggingface.co/docs/transformers/glossary#attention-mask)
* **position\_ids** (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*) — Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0, config.max_position_embeddings - 1]`.

  [What are position IDs?](https://huggingface.co/docs/transformers/glossary#position-ids)
* **head\_mask** (`torch.FloatTensor` of shape `(num_heads,)` or `(num_layers, num_heads)`, *optional*) — Mask to nullify selected heads of the self-attention modules. Mask values selected in `[0, 1]`:
  * 1 indicates the head is **not masked**,
  * 0 indicates the head is **masked**.
* **inputs\_embeds** (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*) — Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert `input_ids` indices into associated vectors than the model’s internal embedding lookup matrix.
* **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.
* **return\_dict** (`bool`, *optional*) — Whether or not to return a [ModelOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.utils.ModelOutput) instead of a plain tuple.
* **labels** (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*) — Labels for computing the masked language modeling loss. Indices should be in `[-100, 0, ..., config.vocab_size]` (see `input_ids` docstring) Tokens with indices set to `-100` are ignored (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`
* **kwargs** (`Dict[str, any]`, optional, defaults to *{}*) — Used to hide legacy arguments that have been deprecated.

Returns

[transformers.modeling\_outputs.MaskedLMOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_outputs.MaskedLMOutput) or `tuple(torch.FloatTensor)`

A [transformers.modeling\_outputs.MaskedLMOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_outputs.MaskedLMOutput) 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 ([EsmConfig](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmConfig)) and inputs.

* **loss** (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided) — Masked language modeling (MLM) loss.
* **logits** (`torch.FloatTensor` of shape `(batch_size, sequence_length, config.vocab_size)`) — Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).
* **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, if the model has an embedding layer, + one for the output of each layer) of shape `(batch_size, sequence_length, hidden_size)`.

  Hidden-states of the model at the output of each layer plus the optional initial embedding outputs.
* **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 after the attention softmax, used to compute the weighted average in the self-attention heads.

The [EsmForMaskedLM](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmForMaskedLM) 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:

Copied

```
>>> from transformers import AutoTokenizer, EsmForMaskedLM
>>> import torch

>>> tokenizer = AutoTokenizer.from_pretrained("facebook/esm2_t6_8M_UR50D")
>>> model = EsmForMaskedLM.from_pretrained("facebook/esm2_t6_8M_UR50D")

>>> inputs = tokenizer("The capital of France is <mask>.", return_tensors="pt")

>>> with torch.no_grad():
...     logits = model(**inputs).logits

>>> # retrieve index of <mask>
>>> mask_token_index = (inputs.input_ids == tokenizer.mask_token_id)[0].nonzero(as_tuple=True)[0]

>>> predicted_token_id = logits[0, mask_token_index].argmax(axis=-1)

>>> labels = tokenizer("The capital of France is Paris.", return_tensors="pt")["input_ids"]
>>> # mask labels of non-<mask> tokens
>>> labels = torch.where(inputs.input_ids == tokenizer.mask_token_id, labels, -100)

>>> outputs = model(**inputs, labels=labels)
```

### EsmForSequenceClassification

#### class transformers.EsmForSequenceClassification

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/modeling_esm.py#L1082)

( config )

Parameters

* **config** ([EsmConfig](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmConfig)) — 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()](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights.

ESM Model transformer with a sequence classification/regression head on top (a linear layer on top of the pooled output) e.g. for GLUE tasks.

This model inherits from [PreTrainedModel](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/model#transformers.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](https://pytorch.org/docs/stable/nn.html#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**

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/modeling_esm.py#L1093)

( input\_ids: typing.Optional\[torch.LongTensor] = Noneattention\_mask: typing.Optional\[torch.Tensor] = Noneposition\_ids: typing.Optional\[torch.LongTensor] = Nonehead\_mask: typing.Optional\[torch.Tensor] = Noneinputs\_embeds: typing.Optional\[torch.FloatTensor] = Nonelabels: typing.Optional\[torch.LongTensor] = Noneoutput\_attentions: typing.Optional\[bool] = Noneoutput\_hidden\_states: typing.Optional\[bool] = Nonereturn\_dict: typing.Optional\[bool] = None ) → [transformers.modeling\_outputs.SequenceClassifierOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_outputs.SequenceClassifierOutput) or `tuple(torch.FloatTensor)`

Parameters

* **input\_ids** (`torch.LongTensor` of shape `(batch_size, sequence_length)`) — Indices of input sequence tokens in the vocabulary.

  Indices can be obtained using [AutoTokenizer](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/tokenizer#transformers.PreTrainedTokenizerFast.encode) and [PreTrainedTokenizer.**call**()](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/vits#transformers.VitsTokenizer.__call__) for details.

  [What are input IDs?](https://huggingface.co/docs/transformers/glossary#input-ids)
* **attention\_mask** (`torch.FloatTensor` 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?](https://huggingface.co/docs/transformers/glossary#attention-mask)
* **position\_ids** (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*) — Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0, config.max_position_embeddings - 1]`.

  [What are position IDs?](https://huggingface.co/docs/transformers/glossary#position-ids)
* **head\_mask** (`torch.FloatTensor` of shape `(num_heads,)` or `(num_layers, num_heads)`, *optional*) — Mask to nullify selected heads of the self-attention modules. Mask values selected in `[0, 1]`:
  * 1 indicates the head is **not masked**,
  * 0 indicates the head is **masked**.
* **inputs\_embeds** (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*) — Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert `input_ids` indices into associated vectors than the model’s internal embedding lookup matrix.
* **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.
* **return\_dict** (`bool`, *optional*) — Whether or not to return a [ModelOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.utils.ModelOutput) instead of a plain tuple.
* **labels** (`torch.LongTensor` of shape `(batch_size,)`, *optional*) — Labels for computing the sequence classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If `config.num_labels > 1` a classification loss is computed (Cross-Entropy).

Returns

[transformers.modeling\_outputs.SequenceClassifierOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_outputs.SequenceClassifierOutput) or `tuple(torch.FloatTensor)`

A [transformers.modeling\_outputs.SequenceClassifierOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_outputs.SequenceClassifierOutput) 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 ([EsmConfig](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmConfig)) and inputs.

* **loss** (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided) — Classification (or regression if config.num\_labels==1) loss.
* **logits** (`torch.FloatTensor` of shape `(batch_size, config.num_labels)`) — Classification (or regression if config.num\_labels==1) scores (before SoftMax).
* **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, if the model has an embedding layer, + one for the output of each layer) of shape `(batch_size, sequence_length, hidden_size)`.

  Hidden-states of the model at the output of each layer plus the optional initial embedding outputs.
* **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 after the attention softmax, used to compute the weighted average in the self-attention heads.

The [EsmForSequenceClassification](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmForSequenceClassification) 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 of single-label classification:

Copied

```
>>> import torch
>>> from transformers import AutoTokenizer, EsmForSequenceClassification

>>> tokenizer = AutoTokenizer.from_pretrained("facebook/esm2_t6_8M_UR50D")
>>> model = EsmForSequenceClassification.from_pretrained("facebook/esm2_t6_8M_UR50D")

>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt")

>>> with torch.no_grad():
...     logits = model(**inputs).logits

>>> predicted_class_id = logits.argmax().item()

>>> # To train a model on `num_labels` classes, you can pass `num_labels=num_labels` to `.from_pretrained(...)`
>>> num_labels = len(model.config.id2label)
>>> model = EsmForSequenceClassification.from_pretrained("facebook/esm2_t6_8M_UR50D", num_labels=num_labels)

>>> labels = torch.tensor([1])
>>> loss = model(**inputs, labels=labels).loss
```

Example of multi-label classification:

Copied

```
>>> import torch
>>> from transformers import AutoTokenizer, EsmForSequenceClassification

>>> tokenizer = AutoTokenizer.from_pretrained("facebook/esm2_t6_8M_UR50D")
>>> model = EsmForSequenceClassification.from_pretrained("facebook/esm2_t6_8M_UR50D", problem_type="multi_label_classification")

>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt")

>>> with torch.no_grad():
...     logits = model(**inputs).logits

>>> predicted_class_ids = torch.arange(0, logits.shape[-1])[torch.sigmoid(logits).squeeze(dim=0) > 0.5]

>>> # To train a model on `num_labels` classes, you can pass `num_labels=num_labels` to `.from_pretrained(...)`
>>> num_labels = len(model.config.id2label)
>>> model = EsmForSequenceClassification.from_pretrained(
...     "facebook/esm2_t6_8M_UR50D", num_labels=num_labels, problem_type="multi_label_classification"
... )

>>> labels = torch.sum(
...     torch.nn.functional.one_hot(predicted_class_ids[None, :].clone(), num_classes=num_labels), dim=1
... ).to(torch.float)
>>> loss = model(**inputs, labels=labels).loss
```

### EsmForTokenClassification

#### class transformers.EsmForTokenClassification

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/modeling_esm.py#L1176)

( config )

Parameters

* **config** ([EsmConfig](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmConfig)) — 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()](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights.

ESM Model with a token classification head on top (a linear layer on top of the hidden-states output) e.g. for Named-Entity-Recognition (NER) tasks.

This model inherits from [PreTrainedModel](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/model#transformers.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](https://pytorch.org/docs/stable/nn.html#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**

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/modeling_esm.py#L1187)

( input\_ids: typing.Optional\[torch.LongTensor] = Noneattention\_mask: typing.Optional\[torch.Tensor] = Noneposition\_ids: typing.Optional\[torch.LongTensor] = Nonehead\_mask: typing.Optional\[torch.Tensor] = Noneinputs\_embeds: typing.Optional\[torch.FloatTensor] = Nonelabels: typing.Optional\[torch.LongTensor] = Noneoutput\_attentions: typing.Optional\[bool] = Noneoutput\_hidden\_states: typing.Optional\[bool] = Nonereturn\_dict: typing.Optional\[bool] = None ) → [transformers.modeling\_outputs.TokenClassifierOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_outputs.TokenClassifierOutput) or `tuple(torch.FloatTensor)`

Parameters

* **input\_ids** (`torch.LongTensor` of shape `(batch_size, sequence_length)`) — Indices of input sequence tokens in the vocabulary.

  Indices can be obtained using [AutoTokenizer](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/tokenizer#transformers.PreTrainedTokenizerFast.encode) and [PreTrainedTokenizer.**call**()](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/vits#transformers.VitsTokenizer.__call__) for details.

  [What are input IDs?](https://huggingface.co/docs/transformers/glossary#input-ids)
* **attention\_mask** (`torch.FloatTensor` 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?](https://huggingface.co/docs/transformers/glossary#attention-mask)
* **position\_ids** (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*) — Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0, config.max_position_embeddings - 1]`.

  [What are position IDs?](https://huggingface.co/docs/transformers/glossary#position-ids)
* **head\_mask** (`torch.FloatTensor` of shape `(num_heads,)` or `(num_layers, num_heads)`, *optional*) — Mask to nullify selected heads of the self-attention modules. Mask values selected in `[0, 1]`:
  * 1 indicates the head is **not masked**,
  * 0 indicates the head is **masked**.
* **inputs\_embeds** (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*) — Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert `input_ids` indices into associated vectors than the model’s internal embedding lookup matrix.
* **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.
* **return\_dict** (`bool`, *optional*) — Whether or not to return a [ModelOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.utils.ModelOutput) instead of a plain tuple.
* **labels** (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*) — Labels for computing the token classification loss. Indices should be in `[0, ..., config.num_labels - 1]`.

Returns

[transformers.modeling\_outputs.TokenClassifierOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_outputs.TokenClassifierOutput) or `tuple(torch.FloatTensor)`

A [transformers.modeling\_outputs.TokenClassifierOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_outputs.TokenClassifierOutput) 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 ([EsmConfig](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmConfig)) and inputs.

* **loss** (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided) — Classification loss.
* **logits** (`torch.FloatTensor` of shape `(batch_size, sequence_length, config.num_labels)`) — Classification scores (before SoftMax).
* **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, if the model has an embedding layer, + one for the output of each layer) of shape `(batch_size, sequence_length, hidden_size)`.

  Hidden-states of the model at the output of each layer plus the optional initial embedding outputs.
* **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 after the attention softmax, used to compute the weighted average in the self-attention heads.

The [EsmForTokenClassification](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmForTokenClassification) 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:

Copied

```
>>> from transformers import AutoTokenizer, EsmForTokenClassification
>>> import torch

>>> tokenizer = AutoTokenizer.from_pretrained("facebook/esm2_t6_8M_UR50D")
>>> model = EsmForTokenClassification.from_pretrained("facebook/esm2_t6_8M_UR50D")

>>> inputs = tokenizer(
...     "BOINCAI is a company based in Paris and New York", add_special_tokens=False, return_tensors="pt"
... )

>>> with torch.no_grad():
...     logits = model(**inputs).logits

>>> predicted_token_class_ids = logits.argmax(-1)

>>> # Note that tokens are classified rather then input words which means that
>>> # there might be more predicted token classes than words.
>>> # Multiple token classes might account for the same word
>>> predicted_tokens_classes = [model.config.id2label[t.item()] for t in predicted_token_class_ids[0]]

>>> labels = predicted_token_class_ids
>>> loss = model(**inputs, labels=labels).loss
```

### EsmForProteinFolding

#### class transformers.EsmForProteinFolding

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/modeling_esmfold.py#L2020)

( config )

Parameters

* **config** ([EsmConfig](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmConfig)) — 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()](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights.

ESMForProteinFolding is the BOINC AI port of the original ESMFold model. It consists of an ESM-2 “stem” followed by a protein folding “head”, although unlike most other output heads, this “head” is similar in size and runtime to the rest of the model combined! It outputs a dictionary containing predicted structural information about the input protein(s).

This model inherits from [PreTrainedModel](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/model#transformers.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](https://pytorch.org/docs/stable/nn.html#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**

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/modeling_esmfold.py#L2084)

( input\_ids: Tensorattention\_mask: typing.Optional\[torch.Tensor] = Noneposition\_ids: typing.Optional\[torch.Tensor] = Nonemasking\_pattern: typing.Optional\[torch.Tensor] = Nonenum\_recycles: typing.Optional\[int] = None ) → `transformers.models.esm.modeling_esmfold.EsmForProteinFoldingOutput` or `tuple(torch.FloatTensor)`

Parameters

* **input\_ids** (`torch.LongTensor` of shape `(batch_size, sequence_length)`) — Indices of input sequence tokens in the vocabulary.

  Indices can be obtained using [AutoTokenizer](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/tokenizer#transformers.PreTrainedTokenizerFast.encode) and [PreTrainedTokenizer.**call**()](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/vits#transformers.VitsTokenizer.__call__) for details.

  [What are input IDs?](https://huggingface.co/docs/transformers/glossary#input-ids)
* **attention\_mask** (`torch.FloatTensor` 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?](https://huggingface.co/docs/transformers/glossary#attention-mask)
* **position\_ids** (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*) — Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0, config.max_position_embeddings - 1]`.

  [What are position IDs?](https://huggingface.co/docs/transformers/glossary#position-ids)
* **masking\_pattern** (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*) — Locations of tokens to mask during training as a form of regularization. Mask values selected in `[0, 1]`.
* **num\_recycles** (`int`, *optional*, defaults to `None`) — Number of times to recycle the input sequence. If `None`, defaults to `config.num_recycles`. “Recycling” consists of passing the output of the folding trunk back in as input to the trunk. During training, the number of recycles should vary with each batch, to ensure that the model learns to output valid predictions after each recycle. During inference, num\_recycles should be set to the highest value that the model was trained with for maximum accuracy. Accordingly, when this value is set to `None`, config.max\_recycles is used.

Returns

`transformers.models.esm.modeling_esmfold.EsmForProteinFoldingOutput` or `tuple(torch.FloatTensor)`

A `transformers.models.esm.modeling_esmfold.EsmForProteinFoldingOutput` 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 (`<class 'transformers.models.esm.configuration_esm.EsmConfig'>`) and inputs.

* **frames** (`torch.FloatTensor`) — Output frames.
* **sidechain\_frames** (`torch.FloatTensor`) — Output sidechain frames.
* **unnormalized\_angles** (`torch.FloatTensor`) — Predicted unnormalized backbone and side chain torsion angles.
* **angles** (`torch.FloatTensor`) — Predicted backbone and side chain torsion angles.
* **positions** (`torch.FloatTensor`) — Predicted positions of the backbone and side chain atoms.
* **states** (`torch.FloatTensor`) — Hidden states from the protein folding trunk.
* **s\_s** (`torch.FloatTensor`) — Per-residue embeddings derived by concatenating the hidden states of each layer of the ESM-2 LM stem.
* **s\_z** (`torch.FloatTensor`) — Pairwise residue embeddings.
* **distogram\_logits** (`torch.FloatTensor`) — Input logits to the distogram used to compute residue distances.
* **lm\_logits** (`torch.FloatTensor`) — Logits output by the ESM-2 protein language model stem.
* **aatype** (`torch.FloatTensor`) — Input amino acids (AlphaFold2 indices).
* **atom14\_atom\_exists** (`torch.FloatTensor`) — Whether each atom exists in the atom14 representation.
* **residx\_atom14\_to\_atom37** (`torch.FloatTensor`) — Mapping between atoms in the atom14 and atom37 representations.
* **residx\_atom37\_to\_atom14** (`torch.FloatTensor`) — Mapping between atoms in the atom37 and atom14 representations.
* **atom37\_atom\_exists** (`torch.FloatTensor`) — Whether each atom exists in the atom37 representation.
* **residue\_index** (`torch.FloatTensor`) — The index of each residue in the protein chain. Unless internal padding tokens are used, this will just be a sequence of integers from 0 to `sequence_length`.
* **lddt\_head** (`torch.FloatTensor`) — Raw outputs from the lddt head used to compute plddt.
* **plddt** (`torch.FloatTensor`) — Per-residue confidence scores. Regions of low confidence may indicate areas where the model’s prediction is uncertain, or where the protein structure is disordered.
* **ptm\_logits** (`torch.FloatTensor`) — Raw logits used for computing ptm.
* **ptm** (`torch.FloatTensor`) — TM-score output representing the model’s high-level confidence in the overall structure.
* **aligned\_confidence\_probs** (`torch.FloatTensor`) — Per-residue confidence scores for the aligned structure.
* **predicted\_aligned\_error** (`torch.FloatTensor`) — Predicted error between the model’s prediction and the ground truth.
* **max\_predicted\_aligned\_error** (`torch.FloatTensor`) — Per-sample maximum predicted error.

The [EsmForProteinFolding](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmForProteinFolding) 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:

Copied

```
>>> from transformers import AutoTokenizer, EsmForProteinFolding

>>> model = EsmForProteinFolding.from_pretrained("facebook/esmfold_v1")
>>> tokenizer = AutoTokenizer.from_pretrained("facebook/esmfold_v1")
>>> inputs = tokenizer(["MLKNVQVQLV"], return_tensors="pt", add_special_tokens=False)  # A tiny random peptide
>>> outputs = model(**inputs)
>>> folded_positions = outputs.positions
```

### TFEsmModel

#### class transformers.TFEsmModel

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/modeling_tf_esm.py#L975)

( \*args\*\*kwargs )

Parameters

* **config** ([EsmConfig](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmConfig)) — 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()](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/model#transformers.TFPreTrainedModel.from_pretrained) method to load the model weights.

The bare ESM Model transformer outputting raw hidden-states without any specific head on top.

This model inherits from [TFPreTrainedModel](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/model#transformers.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 Keras [Model](https://www.tensorflow.org/api_docs/python/tf/keras/Model) subclass. Use it as a regular Keras model and refer to the TF/Keras documentation for all matters related to general usage and behavior.

**call**

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/modeling_tf_esm.py#L981)

( input\_ids: TFModelInputType | None = Noneattention\_mask: np.ndarray | tf.Tensor | None = Noneposition\_ids: np.ndarray | tf.Tensor | None = Nonehead\_mask: np.ndarray | tf.Tensor | None = Noneinputs\_embeds: np.ndarray | tf.Tensor | None = Noneencoder\_hidden\_states: np.ndarray | tf.Tensor | None = Noneencoder\_attention\_mask: np.ndarray | tf.Tensor | None = Nonepast\_key\_values: Optional\[Tuple\[Tuple\[Union\[np.ndarray, tf.Tensor]]]] = Noneuse\_cache: Optional\[bool] = Noneoutput\_attentions: Optional\[bool] = Noneoutput\_hidden\_states: Optional\[bool] = Nonereturn\_dict: Optional\[bool] = Nonetraining: Optional\[bool] = False ) → [transformers.modeling\_tf\_outputs.TFBaseModelOutputWithPoolingAndCrossAttentions](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_tf_outputs.TFBaseModelOutputWithPoolingAndCrossAttentions) or `tuple(tf.Tensor)`

Parameters

* **input\_ids** (`tf.Tensor` of shape `(batch_size, sequence_length)`) — Indices of input sequence tokens in the vocabulary.

  Indices can be obtained using [AutoTokenizer](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/tokenizer#transformers.PreTrainedTokenizerFast.encode) and [PreTrainedTokenizer.**call**()](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/vits#transformers.VitsTokenizer.__call__) for details.

  [What are input IDs?](https://huggingface.co/docs/transformers/glossary#input-ids)
* **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?](https://huggingface.co/docs/transformers/glossary#attention-mask)
* **position\_ids** (`tf.Tensor` of shape `(batch_size, sequence_length)`, *optional*) — Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0, config.max_position_embeddings - 1]`.

  [What are position IDs?](https://huggingface.co/docs/transformers/glossary#position-ids)
* **head\_mask** (`tf.Tensor` of shape `(num_heads,)` or `(num_layers, num_heads)`, *optional*) — Mask to nullify selected heads of the self-attention modules. Mask values selected in `[0, 1]`:
  * 1 indicates the head is **not masked**,
  * 0 indicates the head is **masked**.
* **inputs\_embeds** (`tf.Tensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*) — Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert `input_ids` indices into associated vectors than the model’s internal embedding lookup matrix.
* **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.
* **return\_dict** (`bool`, *optional*) — Whether or not to return a [ModelOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.utils.ModelOutput) instead of a plain tuple.
* **encoder\_hidden\_states** (`tf.Tensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*) — Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention if the model is configured as a decoder.
* **encoder\_attention\_mask** (`tf.Tensor` of shape `(batch_size, sequence_length)`, *optional*) — Mask to avoid performing attention on the padding token indices of the encoder input. This mask is used in the cross-attention if the model is configured as a decoder. Mask values selected in `[0, 1]`:
  * 1 for tokens that are **not masked**,
  * 0 for tokens that are **masked**.
* **past\_key\_values** (`Tuple[Tuple[tf.Tensor]]` of length `config.n_layers`) — contains precomputed key and value hidden states of the attention blocks. Can be used to speed up decoding. If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids` (those that don’t have their past key value states given to this model) of shape `(batch_size, 1)` instead of all `decoder_input_ids` of shape `(batch_size, sequence_length)`.
* **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`). Set to `False` during training, `True` during generation

Returns

[transformers.modeling\_tf\_outputs.TFBaseModelOutputWithPoolingAndCrossAttentions](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_tf_outputs.TFBaseModelOutputWithPoolingAndCrossAttentions) or `tuple(tf.Tensor)`

A [transformers.modeling\_tf\_outputs.TFBaseModelOutputWithPoolingAndCrossAttentions](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_tf_outputs.TFBaseModelOutputWithPoolingAndCrossAttentions) or a tuple of `tf.Tensor` (if `return_dict=False` is passed or when `config.return_dict=False`) comprising various elements depending on the configuration ([EsmConfig](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmConfig)) and inputs.

* **last\_hidden\_state** (`tf.Tensor` of shape `(batch_size, sequence_length, hidden_size)`) — Sequence of hidden-states at the output of the last layer of the model.
* **pooler\_output** (`tf.Tensor` of shape `(batch_size, hidden_size)`) — Last layer hidden-state of the first token of the sequence (classification token) further processed by a Linear layer and a Tanh activation function. The Linear layer weights are trained from the next sentence prediction (classification) objective during pretraining.

  This output is usually *not* a good summary of the semantic content of the input, you’re often better with averaging or pooling the sequence of hidden-states for the whole input sequence.
* **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 pre-computed hidden-states (key and values in the attention blocks) that can be used (see `past_key_values` input) to speed up sequential decoding.
* **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 + one for the output of each layer) of shape `(batch_size, sequence_length, hidden_size)`.

  Hidden-states of the model at the output of each layer plus the initial embedding outputs.
* **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 after the attention softmax, used to compute the weighted average in the self-attention heads.
* **cross\_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 decoder’s cross-attention layer, after the attention softmax, used to compute the weighted average in the cross-attention heads.

The [TFEsmModel](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.TFEsmModel) 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:

Copied

```
>>> from transformers import AutoTokenizer, TFEsmModel
>>> import tensorflow as tf

>>> tokenizer = AutoTokenizer.from_pretrained("facebook/esm2_t6_8M_UR50D")
>>> model = TFEsmModel.from_pretrained("facebook/esm2_t6_8M_UR50D")

>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="tf")
>>> outputs = model(inputs)

>>> last_hidden_states = outputs.last_hidden_state
```

### TFEsmForMaskedLM

#### class transformers.TFEsmForMaskedLM

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/modeling_tf_esm.py#L1046)

( \*args\*\*kwargs )

Parameters

* **config** ([EsmConfig](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmConfig)) — 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()](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/model#transformers.TFPreTrainedModel.from_pretrained) method to load the model weights.

ESM Model with a `language modeling` head on top.

This model inherits from [TFPreTrainedModel](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/model#transformers.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 Keras [Model](https://www.tensorflow.org/api_docs/python/tf/keras/Model) subclass. Use it as a regular Keras model and refer to the TF/Keras documentation for all matters related to general usage and behavior.

**call**

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/modeling_tf_esm.py#L1076)

( input\_ids: TFModelInputType | None = Noneattention\_mask: np.ndarray | tf.Tensor | None = Noneposition\_ids: np.ndarray | tf.Tensor | None = Nonehead\_mask: np.ndarray | tf.Tensor | None = Noneinputs\_embeds: np.ndarray | tf.Tensor | None = Noneencoder\_hidden\_states: np.ndarray | tf.Tensor | None = Noneencoder\_attention\_mask: np.ndarray | tf.Tensor | None = Nonelabels: np.ndarray | tf.Tensor | None = Noneoutput\_attentions: Optional\[bool] = Noneoutput\_hidden\_states: Optional\[bool] = Nonereturn\_dict: Optional\[bool] = Nonetraining: bool = False ) → [transformers.modeling\_tf\_outputs.TFMaskedLMOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_tf_outputs.TFMaskedLMOutput) or `tuple(tf.Tensor)`

Parameters

* **input\_ids** (`tf.Tensor` of shape `(batch_size, sequence_length)`) — Indices of input sequence tokens in the vocabulary.

  Indices can be obtained using [AutoTokenizer](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/tokenizer#transformers.PreTrainedTokenizerFast.encode) and [PreTrainedTokenizer.**call**()](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/vits#transformers.VitsTokenizer.__call__) for details.

  [What are input IDs?](https://huggingface.co/docs/transformers/glossary#input-ids)
* **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?](https://huggingface.co/docs/transformers/glossary#attention-mask)
* **position\_ids** (`tf.Tensor` of shape `(batch_size, sequence_length)`, *optional*) — Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0, config.max_position_embeddings - 1]`.

  [What are position IDs?](https://huggingface.co/docs/transformers/glossary#position-ids)
* **head\_mask** (`tf.Tensor` of shape `(num_heads,)` or `(num_layers, num_heads)`, *optional*) — Mask to nullify selected heads of the self-attention modules. Mask values selected in `[0, 1]`:
  * 1 indicates the head is **not masked**,
  * 0 indicates the head is **masked**.
* **inputs\_embeds** (`tf.Tensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*) — Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert `input_ids` indices into associated vectors than the model’s internal embedding lookup matrix.
* **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.
* **return\_dict** (`bool`, *optional*) — Whether or not to return a [ModelOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.utils.ModelOutput) instead of a plain tuple.
* **labels** (`tf.Tensor` of shape `(batch_size, sequence_length)`, *optional*) — Labels for computing the masked language modeling loss. Indices should be in `[-100, 0, ..., config.vocab_size]` (see `input_ids` docstring) Tokens with indices set to `-100` are ignored (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`
* **kwargs** (`Dict[str, any]`, optional, defaults to *{}*) — Used to hide legacy arguments that have been deprecated.

Returns

[transformers.modeling\_tf\_outputs.TFMaskedLMOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_tf_outputs.TFMaskedLMOutput) or `tuple(tf.Tensor)`

A [transformers.modeling\_tf\_outputs.TFMaskedLMOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_tf_outputs.TFMaskedLMOutput) or a tuple of `tf.Tensor` (if `return_dict=False` is passed or when `config.return_dict=False`) comprising various elements depending on the configuration ([EsmConfig](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmConfig)) and inputs.

* **loss** (`tf.Tensor` of shape `(n,)`, *optional*, where n is the number of non-masked labels, returned when `labels` is provided) — Masked language modeling (MLM) loss.
* **logits** (`tf.Tensor` of shape `(batch_size, sequence_length, config.vocab_size)`) — Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).
* **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 + one for the output of each layer) of shape `(batch_size, sequence_length, hidden_size)`.

  Hidden-states of the model at the output of each layer plus the initial embedding outputs.
* **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 after the attention softmax, used to compute the weighted average in the self-attention heads.

The [TFEsmForMaskedLM](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.TFEsmForMaskedLM) 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:

Copied

```
>>> from transformers import AutoTokenizer, TFEsmForMaskedLM
>>> import tensorflow as tf

>>> tokenizer = AutoTokenizer.from_pretrained("facebook/esm2_t6_8M_UR50D")
>>> model = TFEsmForMaskedLM.from_pretrained("facebook/esm2_t6_8M_UR50D")

>>> inputs = tokenizer("The capital of France is <mask>.", return_tensors="tf")
>>> logits = model(**inputs).logits

>>> # retrieve index of <mask>
>>> mask_token_index = tf.where((inputs.input_ids == tokenizer.mask_token_id)[0])
>>> selected_logits = tf.gather_nd(logits[0], indices=mask_token_index)

>>> predicted_token_id = tf.math.argmax(selected_logits, axis=-1)
```

Copied

```
>>> labels = tokenizer("The capital of France is Paris.", return_tensors="tf")["input_ids"]
>>> # mask labels of non-<mask> tokens
>>> labels = tf.where(inputs.input_ids == tokenizer.mask_token_id, labels, -100)

>>> outputs = model(**inputs, labels=labels)
```

### TFEsmForSequenceClassification

#### class transformers.TFEsmForSequenceClassification

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/modeling_tf_esm.py#L1194)

( \*args\*\*kwargs )

Parameters

* **config** ([EsmConfig](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmConfig)) — 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()](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/model#transformers.TFPreTrainedModel.from_pretrained) method to load the model weights.

ESM Model transformer with a sequence classification/regression head on top (a linear layer on top of the pooled output) e.g. for GLUE tasks.

This model inherits from [TFPreTrainedModel](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/model#transformers.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 Keras [Model](https://www.tensorflow.org/api_docs/python/tf/keras/Model) subclass. Use it as a regular Keras model and refer to the TF/Keras documentation for all matters related to general usage and behavior.

**call**

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/modeling_tf_esm.py#L1205)

( input\_ids: TFModelInputType | None = Noneattention\_mask: np.ndarray | tf.Tensor | None = Noneposition\_ids: np.ndarray | tf.Tensor | None = Nonehead\_mask: np.ndarray | tf.Tensor | None = Noneinputs\_embeds: np.ndarray | tf.Tensor | None = Nonelabels: np.ndarray | tf.Tensor | None = Noneoutput\_attentions: Optional\[bool] = Noneoutput\_hidden\_states: Optional\[bool] = Nonereturn\_dict: Optional\[bool] = Nonetraining: bool = False ) → [transformers.modeling\_tf\_outputs.TFSequenceClassifierOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_tf_outputs.TFSequenceClassifierOutput) or `tuple(tf.Tensor)`

Parameters

* **input\_ids** (`tf.Tensor` of shape `(batch_size, sequence_length)`) — Indices of input sequence tokens in the vocabulary.

  Indices can be obtained using [AutoTokenizer](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/tokenizer#transformers.PreTrainedTokenizerFast.encode) and [PreTrainedTokenizer.**call**()](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/vits#transformers.VitsTokenizer.__call__) for details.

  [What are input IDs?](https://huggingface.co/docs/transformers/glossary#input-ids)
* **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?](https://huggingface.co/docs/transformers/glossary#attention-mask)
* **position\_ids** (`tf.Tensor` of shape `(batch_size, sequence_length)`, *optional*) — Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0, config.max_position_embeddings - 1]`.

  [What are position IDs?](https://huggingface.co/docs/transformers/glossary#position-ids)
* **head\_mask** (`tf.Tensor` of shape `(num_heads,)` or `(num_layers, num_heads)`, *optional*) — Mask to nullify selected heads of the self-attention modules. Mask values selected in `[0, 1]`:
  * 1 indicates the head is **not masked**,
  * 0 indicates the head is **masked**.
* **inputs\_embeds** (`tf.Tensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*) — Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert `input_ids` indices into associated vectors than the model’s internal embedding lookup matrix.
* **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.
* **return\_dict** (`bool`, *optional*) — Whether or not to return a [ModelOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.utils.ModelOutput) instead of a plain tuple.
* **labels** (`tf.Tensor` of shape `(batch_size,)`, *optional*) — Labels for computing the sequence classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If `config.num_labels > 1` a classification loss is computed (Cross-Entropy).

Returns

[transformers.modeling\_tf\_outputs.TFSequenceClassifierOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_tf_outputs.TFSequenceClassifierOutput) or `tuple(tf.Tensor)`

A [transformers.modeling\_tf\_outputs.TFSequenceClassifierOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_tf_outputs.TFSequenceClassifierOutput) or a tuple of `tf.Tensor` (if `return_dict=False` is passed or when `config.return_dict=False`) comprising various elements depending on the configuration ([EsmConfig](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmConfig)) and inputs.

* **loss** (`tf.Tensor` of shape `(batch_size, )`, *optional*, returned when `labels` is provided) — Classification (or regression if config.num\_labels==1) loss.
* **logits** (`tf.Tensor` of shape `(batch_size, config.num_labels)`) — Classification (or regression if config.num\_labels==1) scores (before SoftMax).
* **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 + one for the output of each layer) of shape `(batch_size, sequence_length, hidden_size)`.

  Hidden-states of the model at the output of each layer plus the initial embedding outputs.
* **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 after the attention softmax, used to compute the weighted average in the self-attention heads.

The [TFEsmForSequenceClassification](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.TFEsmForSequenceClassification) 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:

Copied

```
>>> from transformers import AutoTokenizer, TFEsmForSequenceClassification
>>> import tensorflow as tf

>>> tokenizer = AutoTokenizer.from_pretrained("facebook/esm2_t6_8M_UR50D")
>>> model = TFEsmForSequenceClassification.from_pretrained("facebook/esm2_t6_8M_UR50D")

>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="tf")

>>> logits = model(**inputs).logits

>>> predicted_class_id = int(tf.math.argmax(logits, axis=-1)[0])
```

Copied

```
>>> # To train a model on `num_labels` classes, you can pass `num_labels=num_labels` to `.from_pretrained(...)`
>>> num_labels = len(model.config.id2label)
>>> model = TFEsmForSequenceClassification.from_pretrained("facebook/esm2_t6_8M_UR50D", num_labels=num_labels)

>>> labels = tf.constant(1)
>>> loss = model(**inputs, labels=labels).loss
```

### TFEsmForTokenClassification

#### class transformers.TFEsmForTokenClassification

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/modeling_tf_esm.py#L1268)

( \*args\*\*kwargs )

Parameters

* **config** ([EsmConfig](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmConfig)) — 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()](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/model#transformers.TFPreTrainedModel.from_pretrained) method to load the model weights.

ESM Model with a token classification head on top (a linear layer on top of the hidden-states output) e.g. for Named-Entity-Recognition (NER) tasks.

This model inherits from [TFPreTrainedModel](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/model#transformers.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 Keras [Model](https://www.tensorflow.org/api_docs/python/tf/keras/Model) subclass. Use it as a regular Keras model and refer to the TF/Keras documentation for all matters related to general usage and behavior.

**call**

[\<source>](https://github.com/huggingface/transformers/blob/v4.34.1/src/transformers/models/esm/modeling_tf_esm.py#L1280)

( input\_ids: TFModelInputType | None = Noneattention\_mask: np.ndarray | tf.Tensor | None = Noneposition\_ids: np.ndarray | tf.Tensor | None = Nonehead\_mask: np.ndarray | tf.Tensor | None = Noneinputs\_embeds: np.ndarray | tf.Tensor | None = Nonelabels: np.ndarray | tf.Tensor | None = Noneoutput\_attentions: Optional\[bool] = Noneoutput\_hidden\_states: Optional\[bool] = Nonereturn\_dict: Optional\[bool] = Nonetraining: bool = False ) → [transformers.modeling\_tf\_outputs.TFTokenClassifierOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_tf_outputs.TFTokenClassifierOutput) or `tuple(tf.Tensor)`

Parameters

* **input\_ids** (`tf.Tensor` of shape `(batch_size, sequence_length)`) — Indices of input sequence tokens in the vocabulary.

  Indices can be obtained using [AutoTokenizer](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/tokenizer#transformers.PreTrainedTokenizerFast.encode) and [PreTrainedTokenizer.**call**()](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/vits#transformers.VitsTokenizer.__call__) for details.

  [What are input IDs?](https://huggingface.co/docs/transformers/glossary#input-ids)
* **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?](https://huggingface.co/docs/transformers/glossary#attention-mask)
* **position\_ids** (`tf.Tensor` of shape `(batch_size, sequence_length)`, *optional*) — Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0, config.max_position_embeddings - 1]`.

  [What are position IDs?](https://huggingface.co/docs/transformers/glossary#position-ids)
* **head\_mask** (`tf.Tensor` of shape `(num_heads,)` or `(num_layers, num_heads)`, *optional*) — Mask to nullify selected heads of the self-attention modules. Mask values selected in `[0, 1]`:
  * 1 indicates the head is **not masked**,
  * 0 indicates the head is **masked**.
* **inputs\_embeds** (`tf.Tensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*) — Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert `input_ids` indices into associated vectors than the model’s internal embedding lookup matrix.
* **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.
* **return\_dict** (`bool`, *optional*) — Whether or not to return a [ModelOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.utils.ModelOutput) instead of a plain tuple.
* **labels** (`tf.Tensor` of shape `(batch_size, sequence_length)`, *optional*) — Labels for computing the token classification loss. Indices should be in `[0, ..., config.num_labels - 1]`.

Returns

[transformers.modeling\_tf\_outputs.TFTokenClassifierOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_tf_outputs.TFTokenClassifierOutput) or `tuple(tf.Tensor)`

A [transformers.modeling\_tf\_outputs.TFTokenClassifierOutput](https://huggingface.co/docs/transformers/v4.34.1/en/main_classes/output#transformers.modeling_tf_outputs.TFTokenClassifierOutput) or a tuple of `tf.Tensor` (if `return_dict=False` is passed or when `config.return_dict=False`) comprising various elements depending on the configuration ([EsmConfig](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.EsmConfig)) and inputs.

* **loss** (`tf.Tensor` of shape `(n,)`, *optional*, where n is the number of unmasked labels, returned when `labels` is provided) — Classification loss.
* **logits** (`tf.Tensor` of shape `(batch_size, sequence_length, config.num_labels)`) — Classification scores (before SoftMax).
* **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 + one for the output of each layer) of shape `(batch_size, sequence_length, hidden_size)`.

  Hidden-states of the model at the output of each layer plus the initial embedding outputs.
* **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 after the attention softmax, used to compute the weighted average in the self-attention heads.

The [TFEsmForTokenClassification](https://huggingface.co/docs/transformers/v4.34.1/en/model_doc/esm#transformers.TFEsmForTokenClassification) 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:

Copied

```
>>> from transformers import AutoTokenizer, TFEsmForTokenClassification
>>> import tensorflow as tf

>>> tokenizer = AutoTokenizer.from_pretrained("facebook/esm2_t6_8M_UR50D")
>>> model = TFEsmForTokenClassification.from_pretrained("facebook/esm2_t6_8M_UR50D")

>>> inputs = tokenizer(
...     "BOINCAI is a company based in Paris and New York", add_special_tokens=False, return_tensors="tf"
... )

>>> logits = model(**inputs).logits
>>> predicted_token_class_ids = tf.math.argmax(logits, axis=-1)

>>> # Note that tokens are classified rather then input words which means that
>>> # there might be more predicted token classes than words.
>>> # Multiple token classes might account for the same word
>>> predicted_tokens_classes = [model.config.id2label[t] for t in predicted_token_class_ids[0].numpy().tolist()]
```

Copied

```
>>> labels = predicted_token_class_ids
>>> loss = tf.math.reduce_mean(model(**inputs, labels=labels).loss)
```


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://boinc-ai.gitbook.io/transformers/api/models/text-models/esm.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.