# UltraBERT

## Overview

UltraBERT was introduced in [UltraBERT: Beyond the Limit of BERT Pre-Training](https://huggingface.co/papers/2610.15028) by Jungwoo Park, Taewhoo Lee, and Jaewoo Kang. It is a family of bidirectional text encoders trained with a combination of masking ratios, pruning, and distillation for classification and retrieval tasks.

The released encoders support sequences of up to 32,768 tokens. They use rotary position embeddings (RoPE), RMSNorm, SwiGLU feed-forward layers, and alternating local and global attention. Every third layer uses global attention. The other layers attend to the current token and up to 128 tokens on either side. The RoPE base is 10,000 for local layers and 2,560,000 for global layers.

| Checkpoint | Layers | Hidden size | Attention heads | Use |
|---|---|---|---|---|
| [UltraBERT-150M](https://huggingface.co/dmis-lab/UltraBERT-150M) | 24 | 768 | 6 | Masked language modeling and downstream fine-tuning |
| [UltraBERT-380M](https://huggingface.co/dmis-lab/UltraBERT-380M) | 24 | 1,024 | 8 | Masked language modeling and downstream fine-tuning |
| [UltraBERT-Embed-150M](https://huggingface.co/dmis-lab/UltraBERT-Embed-150M) | 24 | 768 | 6 | Text embeddings |

The masked language modeling checkpoints contain the prediction head trained with the low masking ratio. Its output weights are untied from the input embeddings. The tokenizer uses the SmolLM2 vocabulary of 49,152 tokens.

## Usage examples

### Masked language modeling

Use [Pipeline](/docs/transformers/v5.20.0/en/main_classes/pipelines#transformers.Pipeline) or [AutoModelForMaskedLM](/docs/transformers/v5.20.0/en/model_doc/auto#transformers.AutoModelForMaskedLM) to predict a masked token. The mask token is `<|mask|>`.

```python
from transformers import pipeline

fill_mask = pipeline(
    task="fill-mask",
    model="dmis-lab/UltraBERT-150M",
    device_map="auto",
)
print(fill_mask("The capital of France is <|mask|>."))
# [
#     {'score': 0.9612365961074829, 'token': 7042, 'token_str': ' Paris',
#      'sequence': 'The capital of France is Paris.'},
#     {'score': 0.00930546224117279, 'token': 38034, 'token_str': ' Lyon',
#      'sequence': 'The capital of France is Lyon.'},
#     {'score': 0.008486305363476276, 'token': 44558, 'token_str': ' Nice',
#      'sequence': 'The capital of France is Nice.'},
#     {'score': 0.0034405170008540154, 'token': 37628, 'token_str': ' Versailles',
#      'sequence': 'The capital of France is Versailles.'},
#     {'score': 0.0014691735850647092, 'token': 4649, 'token_str': ' France',
#      'sequence': 'The capital of France is France.'}
# ]
```

```python
import torch
from transformers import AutoModelForMaskedLM, AutoTokenizer

checkpoint = "dmis-lab/UltraBERT-150M"
tokenizer = AutoTokenizer.from_pretrained(checkpoint)
model = AutoModelForMaskedLM.from_pretrained(checkpoint, device_map="auto")

inputs = tokenizer("The capital of France is <|mask|>.", return_tensors="pt").to(model.device)
with torch.no_grad():
    logits = model(**inputs).logits

masked_logits = logits[inputs.input_ids == tokenizer.mask_token_id]
print(tokenizer.batch_decode(masked_logits.argmax(dim=-1)[:, None]))
# [' Paris']
```

### Text embeddings

UltraBERT-Embed-150M uses the first token's hidden state as the text embedding. For retrieval, prepend the task instruction to queries and encode documents without an instruction. Use right padding to preserve the first token's position. See the [model card](https://huggingface.co/dmis-lab/UltraBERT-Embed-150M) for instructions for other tasks.

The examples below select the first token's hidden state and normalize the embeddings so their dot product gives cosine similarity.

```python
import torch
import torch.nn.functional as F
from transformers import pipeline

extractor = pipeline(
    task="feature-extraction",
    model="dmis-lab/UltraBERT-Embed-150M",
    device_map="auto",
)

texts = [
    "Instruct: Given a question, retrieve passages that answer the question\nQuery: What is the capital of France?",
    "Paris is the capital of France.",
    "Berlin is the capital of Germany.",
]
features = extractor(texts, return_tensors=True, truncation=True, tokenize_kwargs={"max_length": 32768})
embeddings = F.normalize(torch.cat([output[:, 0] for output in features]), p=2, dim=-1)

print(embeddings[:1] @ embeddings[1:].T)
# tensor([[0.6527, 0.3981]])
```

```python
import torch
import torch.nn.functional as F
from transformers import AutoModel, AutoTokenizer

checkpoint = "dmis-lab/UltraBERT-Embed-150M"
tokenizer = AutoTokenizer.from_pretrained(checkpoint, padding_side="right")
model = AutoModel.from_pretrained(checkpoint, device_map="auto")

texts = [
    "Instruct: Given a question, retrieve passages that answer the question\nQuery: What is the capital of France?",
    "Paris is the capital of France.",
    "Berlin is the capital of Germany.",
]
inputs = tokenizer(texts, padding=True, truncation=True, max_length=32768, return_tensors="pt").to(model.device)
with torch.no_grad():
    embeddings = F.normalize(model(**inputs).last_hidden_state[:, 0], p=2, dim=-1)

print(embeddings[:1] @ embeddings[1:].T)
# tensor([[0.6527, 0.3981]], device='cuda:0')
```

## UltrabertConfig[[transformers.UltrabertConfig]]

#### transformers.UltrabertConfig[[transformers.UltrabertConfig]]

```python
transformers.UltrabertConfig(transformers_version: str | None = None, architectures: list[str] | None = None, output_hidden_states: bool | None = False, return_dict: bool | None = True, dtype: str | torch.dtype | None = None, chunk_size_feed_forward: int = 0, is_encoder_decoder: bool = False, id2label: dict[int, str] | dict[str, str] | None = None, label2id: dict[str, int] | dict[str, str] | None = None, problem_type: Literal['regression', 'single_label_classification', 'multi_label_classification'] | None = None, vocab_size: int = 32000, hidden_size: int = 4096, intermediate_size: int = 11008, num_hidden_layers: int = 32, num_attention_heads: int = 32, num_key_value_heads: int | None = None, hidden_act: str = 'silu', max_position_embeddings: int = 32768, initializer_range: float = 0.02, rms_norm_eps: float = 1e-06, use_cache: bool = False, pad_token_id: int | None = 0, bos_token_id: int | None = 0, eos_token_id: int | list[int] | None = 0, tie_word_embeddings: bool = False, rope_parameters: transformers.modeling_rope_utils.RopeParameters | dict | None = None, attention_bias: bool = False, attention_dropout: int | float = 0.0, mlp_bias: bool = False, head_dim: int | None = None, classifier_pooling: typing.Literal['cls', 'mean'] = 'cls', sliding_window: int = 128, layer_types: list[str] | None = None, prediction_head_size: int = 768, classifier_dropout: float = 0.0)
```

[Source](https://github.com/huggingface/transformers/blob/v5.20.0/src/transformers/models/ultrabert/configuration_ultrabert.py#L30)

**Parameters:**

vocab_size (`int`, *optional*, defaults to `32000`) : Vocabulary size of the model. Defines the number of different tokens that can be represented by the `input_ids`.

hidden_size (`int`, *optional*, defaults to `4096`) : Dimension of the hidden representations.

intermediate_size (`int`, *optional*, defaults to `11008`) : Dimension of the MLP representations.

num_hidden_layers (`int`, *optional*, defaults to `32`) : Number of hidden layers in the Transformer decoder.

num_attention_heads (`int`, *optional*, defaults to `32`) : Number of attention heads for each attention layer in the Transformer decoder.

num_key_value_heads (`int`, *optional*) : This is the number of key_value heads that should be used to implement Grouped Query Attention. If `num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if `num_key_value_heads=1` the model will use Multi Query Attention (MQA) otherwise GQA is used. When converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed by meanpooling all the original heads within that group. For more details, check out [this paper](https://huggingface.co/papers/2305.13245). If it is not specified, will default to `num_attention_heads`.

hidden_act (`str`, *optional*, defaults to `silu`) : The non-linear activation function (function or string) in the decoder. For example, `"gelu"`, `"relu"`, `"silu"`, etc.

max_position_embeddings (`int`, *optional*, defaults to `32768`) : The maximum sequence length that this model might ever be used with.

initializer_range (`float`, *optional*, defaults to `0.02`) : The standard deviation of the truncated_normal_initializer for initializing all weight matrices.

rms_norm_eps (`float`, *optional*, defaults to `1e-06`) : The epsilon used by the rms normalization layers.

use_cache (`bool`, *optional*, defaults to `False`) : Whether or not the model should return the last key/values attentions (not used by all models). Only relevant if `config.is_decoder=True` or when the model is a decoder-only generative model.

pad_token_id (`int`, *optional*, defaults to `0`) : Token id used for padding in the vocabulary.

bos_token_id (`int`, *optional*, defaults to `0`) : Token id used for beginning-of-stream in the vocabulary.

eos_token_id (`Union[int, list[int]]`, *optional*, defaults to `0`) : Token id used for end-of-stream in the vocabulary.

tie_word_embeddings (`bool`, *optional*, defaults to `False`) : Whether to tie weight embeddings according to model's `tied_weights_keys` mapping.

rope_parameters (`Union[~modeling_rope_utils.RopeParameters, dict]`, *optional*) : Dictionary containing the configuration parameters for the RoPE embeddings. The dictionary should contain a value for `rope_theta` and optionally parameters used for scaling in case you want to use RoPE with longer `max_position_embeddings`.

attention_bias (`bool`, *optional*, defaults to `False`) : Whether to use a bias in the query, key, value and output projection layers during self-attention.

attention_dropout (`Union[int, float]`, *optional*, defaults to `0.0`) : The dropout ratio for the attention probabilities.

mlp_bias (`bool`, *optional*, defaults to `False`) : Whether to use a bias in up_proj, down_proj and gate_proj layers in the MLP layers.

head_dim (`int`, *optional*) : The attention head dimension. If None, it will default to hidden_size // num_attention_heads

classifier_pooling (`str`, *optional*, defaults to `"cls"`) : Pooling method for sequence classification. Either `"cls"` or `"mean"`.

sliding_window (`int`, *optional*, defaults to 128) : Inclusive attention radius on either side of each token in sliding attention layers.

layer_types (`list[str]`, *optional*) : A list that explicitly maps each layer index with its layer type. If not provided, it will be automatically generated based on config values.

prediction_head_size (`int`, *optional*, defaults to 768) : Intermediate dimension of the two-layer masked language modeling prediction head.

classifier_dropout (`float`, *optional*, defaults to `0.0`) : The dropout ratio for classifier.

This is the configuration class to store the configuration of a UltrabertModel. It is used to instantiate a Ultrabert
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 [dmis-lab/UltraBERT-150M](https://huggingface.co/dmis-lab/UltraBERT-150M)

Configuration objects inherit from [PreTrainedConfig](/docs/transformers/v5.20.0/en/main_classes/configuration#transformers.PreTrainedConfig) and can be used to control the model outputs. Read the
documentation from [PreTrainedConfig](/docs/transformers/v5.20.0/en/main_classes/configuration#transformers.PreTrainedConfig) for more information.

## UltrabertModel[[transformers.UltrabertModel]]

#### transformers.UltrabertModel[[transformers.UltrabertModel]]

```python
transformers.UltrabertModel(config: UltrabertConfig)
```

[Source](https://github.com/huggingface/transformers/blob/v5.20.0/src/transformers/models/ultrabert/modeling_ultrabert.py#L356)

**Parameters:**

config ([UltrabertConfig](/docs/transformers/v5.20.0/en/model_doc/ultrabert#transformers.UltrabertConfig)) : 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()](/docs/transformers/v5.20.0/en/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights.

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

This model inherits from [PreTrainedModel](/docs/transformers/v5.20.0/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[[transformers.UltrabertModel.forward]]

```python
forward(input_ids: typing.Optional[torch.Tensor] = None, attention_mask: typing.Optional[torch.Tensor] = None, position_ids: typing.Optional[torch.Tensor] = None, inputs_embeds: typing.Optional[torch.Tensor] = None, **kwargs: *<class 'transformers.utils.generic.TransformersKwargs'>)
```

[Source](https://github.com/huggingface/transformers/blob/v5.20.0/src/transformers/models/ultrabert/modeling_ultrabert.py#L373)

**Parameters:**

input_ids (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*) : Indices of input sequence tokens in the vocabulary. Padding will be ignored by default.  Indices can be obtained using [AutoTokenizer](/docs/transformers/v5.20.0/en/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/v5.20.0/en/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and [PreTrainedTokenizer.__call__()](/docs/transformers/v5.20.0/en/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.__call__) for details.  [What are input IDs?](../glossary#input-ids)

attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*) : Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:  - 1 for tokens that are **not masked**, - 0 for tokens that are **masked**.  [What are attention masks?](../glossary#attention-mask)

position_ids (`torch.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.n_positions - 1]`.  [What are position IDs?](../glossary#position-ids)

inputs_embeds (`torch.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.

**Returns:** [BaseModelOutput](/docs/transformers/v5.20.0/en/main_classes/output#transformers.modeling_outputs.BaseModelOutput) or `tuple(torch.FloatTensor)`

A [BaseModelOutput](/docs/transformers/v5.20.0/en/main_classes/output#transformers.modeling_outputs.BaseModelOutput) 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 ([UltrabertConfig](/docs/transformers/v5.20.0/en/model_doc/ultrabert#transformers.UltrabertConfig)) and inputs.

The [UltrabertModel](/docs/transformers/v5.20.0/en/model_doc/ultrabert#transformers.UltrabertModel) 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.

- **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.
- **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.

## UltrabertForMaskedLM[[transformers.UltrabertForMaskedLM]]

#### transformers.UltrabertForMaskedLM[[transformers.UltrabertForMaskedLM]]

```python
transformers.UltrabertForMaskedLM(config: UltrabertConfig)
```

[Source](https://github.com/huggingface/transformers/blob/v5.20.0/src/transformers/models/ultrabert/modeling_ultrabert.py#L437)

**Parameters:**

config ([UltrabertConfig](/docs/transformers/v5.20.0/en/model_doc/ultrabert#transformers.UltrabertConfig)) : 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()](/docs/transformers/v5.20.0/en/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights.

The Ultrabert Model with a decoder head on top that is used for masked language modeling.

This model inherits from [PreTrainedModel](/docs/transformers/v5.20.0/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[[transformers.UltrabertForMaskedLM.forward]]

```python
forward(input_ids: typing.Optional[torch.LongTensor] = None, attention_mask: typing.Optional[torch.Tensor] = None, position_ids: typing.Optional[torch.Tensor] = None, inputs_embeds: typing.Optional[torch.Tensor] = None, labels: typing.Optional[torch.Tensor] = None, **kwargs: *<class 'transformers.utils.generic.TransformersKwargs'>)
```

[Source](https://github.com/huggingface/transformers/blob/v5.20.0/src/transformers/models/ultrabert/modeling_ultrabert.py#L460)

**Parameters:**

input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*) : Indices of input sequence tokens in the vocabulary. Padding will be ignored by default.  Indices can be obtained using [AutoTokenizer](/docs/transformers/v5.20.0/en/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/v5.20.0/en/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and [PreTrainedTokenizer.__call__()](/docs/transformers/v5.20.0/en/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.__call__) for details.  [What are input IDs?](../glossary#input-ids)

attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*) : Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:  - 1 for tokens that are **not masked**, - 0 for tokens that are **masked**.  [What are attention masks?](../glossary#attention-mask)

position_ids (`torch.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.n_positions - 1]`.  [What are position IDs?](../glossary#position-ids)

inputs_embeds (`torch.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.

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]`.

**Returns:** [MaskedLMOutput](/docs/transformers/v5.20.0/en/main_classes/output#transformers.modeling_outputs.MaskedLMOutput) or `tuple(torch.FloatTensor)`

A [MaskedLMOutput](/docs/transformers/v5.20.0/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 ([UltrabertConfig](/docs/transformers/v5.20.0/en/model_doc/ultrabert#transformers.UltrabertConfig)) and inputs.

The [UltrabertForMaskedLM](/docs/transformers/v5.20.0/en/model_doc/ultrabert#transformers.UltrabertForMaskedLM) 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.

- **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.

Example:

```python
>>> from transformers import AutoTokenizer, UltrabertForMaskedLM
>>> import torch

>>> tokenizer = AutoTokenizer.from_pretrained("dmis-lab/UltraBERT-150M")
>>> model = UltrabertForMaskedLM.from_pretrained("dmis-lab/UltraBERT-150M")

>>> 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)
>>> tokenizer.decode(predicted_token_id)
...

>>> 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)
>>> round(outputs.loss.item(), 2)
...
```

## UltrabertForSequenceClassification[[transformers.UltrabertForSequenceClassification]]

#### transformers.UltrabertForSequenceClassification[[transformers.UltrabertForSequenceClassification]]

```python
transformers.UltrabertForSequenceClassification(config: UltrabertConfig)
```

[Source](https://github.com/huggingface/transformers/blob/v5.20.0/src/transformers/models/ultrabert/modeling_ultrabert.py#L509)

**Parameters:**

config ([UltrabertConfig](/docs/transformers/v5.20.0/en/model_doc/ultrabert#transformers.UltrabertConfig)) : 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()](/docs/transformers/v5.20.0/en/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights.

The Ultrabert Model with a sequence classification head on top that performs pooling.

This model inherits from [PreTrainedModel](/docs/transformers/v5.20.0/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[[transformers.UltrabertForSequenceClassification.forward]]

```python
forward(input_ids: typing.Optional[torch.LongTensor] = None, attention_mask: typing.Optional[torch.Tensor] = None, position_ids: typing.Optional[torch.Tensor] = None, inputs_embeds: typing.Optional[torch.Tensor] = None, labels: typing.Optional[torch.Tensor] = None, **kwargs: *<class 'transformers.utils.generic.TransformersKwargs'>)
```

[Source](https://github.com/huggingface/transformers/blob/v5.20.0/src/transformers/models/ultrabert/modeling_ultrabert.py#L523)

**Parameters:**

input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*) : Indices of input sequence tokens in the vocabulary. Padding will be ignored by default.  Indices can be obtained using [AutoTokenizer](/docs/transformers/v5.20.0/en/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/v5.20.0/en/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and [PreTrainedTokenizer.__call__()](/docs/transformers/v5.20.0/en/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.__call__) for details.  [What are input IDs?](../glossary#input-ids)

attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*) : Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:  - 1 for tokens that are **not masked**, - 0 for tokens that are **masked**.  [What are attention masks?](../glossary#attention-mask)

position_ids (`torch.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.n_positions - 1]`.  [What are position IDs?](../glossary#position-ids)

inputs_embeds (`torch.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.

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:** [SequenceClassifierOutput](/docs/transformers/v5.20.0/en/main_classes/output#transformers.modeling_outputs.SequenceClassifierOutput) or `tuple(torch.FloatTensor)`

A [SequenceClassifierOutput](/docs/transformers/v5.20.0/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 ([UltrabertConfig](/docs/transformers/v5.20.0/en/model_doc/ultrabert#transformers.UltrabertConfig)) and inputs.

The [UltrabertForSequenceClassification](/docs/transformers/v5.20.0/en/model_doc/ultrabert#transformers.UltrabertForSequenceClassification) 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.

- **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.

Example of single-label classification:

```python
>>> import torch
>>> from transformers import AutoTokenizer, UltrabertForSequenceClassification

>>> tokenizer = AutoTokenizer.from_pretrained("dmis-lab/UltraBERT-150M")
>>> model = UltrabertForSequenceClassification.from_pretrained("dmis-lab/UltraBERT-150M")

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

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

>>> predicted_class_id = logits.argmax().item()
>>> model.config.id2label[predicted_class_id]
...

>>> # 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 = UltrabertForSequenceClassification.from_pretrained("dmis-lab/UltraBERT-150M", num_labels=num_labels)

>>> labels = torch.tensor([1])
>>> loss = model(**inputs, labels=labels).loss
>>> round(loss.item(), 2)
...
```

Example of multi-label classification:

```python
>>> import torch
>>> from transformers import AutoTokenizer, UltrabertForSequenceClassification

>>> tokenizer = AutoTokenizer.from_pretrained("dmis-lab/UltraBERT-150M")
>>> model = UltrabertForSequenceClassification.from_pretrained("dmis-lab/UltraBERT-150M", 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 = UltrabertForSequenceClassification.from_pretrained(
...     "dmis-lab/UltraBERT-150M", 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
```

## UltrabertForTokenClassification[[transformers.UltrabertForTokenClassification]]

#### transformers.UltrabertForTokenClassification[[transformers.UltrabertForTokenClassification]]

```python
transformers.UltrabertForTokenClassification(config: UltrabertConfig)
```

[Source](https://github.com/huggingface/transformers/blob/v5.20.0/src/transformers/models/ultrabert/modeling_ultrabert.py#L594)

**Parameters:**

config ([UltrabertConfig](/docs/transformers/v5.20.0/en/model_doc/ultrabert#transformers.UltrabertConfig)) : 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()](/docs/transformers/v5.20.0/en/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights.

The Ultrabert Model with a token classification head on top, e.g. for Named Entity Recognition (NER) tasks.

This model inherits from [PreTrainedModel](/docs/transformers/v5.20.0/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[[transformers.UltrabertForTokenClassification.forward]]

```python
forward(input_ids: typing.Optional[torch.LongTensor] = None, attention_mask: typing.Optional[torch.Tensor] = None, position_ids: typing.Optional[torch.Tensor] = None, inputs_embeds: typing.Optional[torch.Tensor] = None, labels: typing.Optional[torch.Tensor] = None, **kwargs: *<class 'transformers.utils.generic.TransformersKwargs'>)
```

[Source](https://github.com/huggingface/transformers/blob/v5.20.0/src/transformers/models/ultrabert/modeling_ultrabert.py#L607)

**Parameters:**

input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*) : Indices of input sequence tokens in the vocabulary. Padding will be ignored by default.  Indices can be obtained using [AutoTokenizer](/docs/transformers/v5.20.0/en/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/v5.20.0/en/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and [PreTrainedTokenizer.__call__()](/docs/transformers/v5.20.0/en/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.__call__) for details.  [What are input IDs?](../glossary#input-ids)

attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*) : Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:  - 1 for tokens that are **not masked**, - 0 for tokens that are **masked**.  [What are attention masks?](../glossary#attention-mask)

position_ids (`torch.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.n_positions - 1]`.  [What are position IDs?](../glossary#position-ids)

inputs_embeds (`torch.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.

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:** [TokenClassifierOutput](/docs/transformers/v5.20.0/en/main_classes/output#transformers.modeling_outputs.TokenClassifierOutput) or `tuple(torch.FloatTensor)`

A [TokenClassifierOutput](/docs/transformers/v5.20.0/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 ([UltrabertConfig](/docs/transformers/v5.20.0/en/model_doc/ultrabert#transformers.UltrabertConfig)) and inputs.

The [UltrabertForTokenClassification](/docs/transformers/v5.20.0/en/model_doc/ultrabert#transformers.UltrabertForTokenClassification) 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.

- **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.

Example:

```python
>>> from transformers import AutoTokenizer, UltrabertForTokenClassification
>>> import torch

>>> tokenizer = AutoTokenizer.from_pretrained("dmis-lab/UltraBERT-150M")
>>> model = UltrabertForTokenClassification.from_pretrained("dmis-lab/UltraBERT-150M")

>>> inputs = tokenizer(
...     "HuggingFace 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]]
>>> predicted_tokens_classes
...

>>> labels = predicted_token_class_ids
>>> loss = model(**inputs, labels=labels).loss
>>> round(loss.item(), 2)
...
```

## UltrabertForQuestionAnswering[[transformers.UltrabertForQuestionAnswering]]

#### transformers.UltrabertForQuestionAnswering[[transformers.UltrabertForQuestionAnswering]]

```python
transformers.UltrabertForQuestionAnswering(config)
```

[Source](https://github.com/huggingface/transformers/blob/v5.20.0/src/transformers/models/ultrabert/modeling_ultrabert.py#L644)

#### forward[[transformers.UltrabertForQuestionAnswering.forward]]

```python
forward(input_ids: torch.LongTensor | None = None, attention_mask: torch.Tensor | None = None, position_ids: torch.LongTensor | None = None, past_key_values: Cache | None = None, inputs_embeds: torch.FloatTensor | None = None, start_positions: torch.LongTensor | None = None, end_positions: torch.LongTensor | None = None, **kwargs: Unpack[TransformersKwargs])
```

[Source](https://github.com/huggingface/transformers/blob/v5.20.0/src/transformers/modeling_layers.py#L208)

**Parameters:**

input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*) : Indices of input sequence tokens in the vocabulary. Padding will be ignored by default.  Indices can be obtained using [AutoTokenizer](/docs/transformers/v5.20.0/en/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/v5.20.0/en/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and [PreTrainedTokenizer.__call__()](/docs/transformers/v5.20.0/en/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.__call__) for details.  [What are input IDs?](../glossary#input-ids)

attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*) : Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:  - 1 for tokens that are **not masked**, - 0 for tokens that are **masked**.  [What are attention masks?](../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.n_positions - 1]`.  [What are position IDs?](../glossary#position-ids)

past_key_values (`Cache`, *optional*) : Pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention blocks) that can be used to speed up sequential decoding. This typically consists in the `past_key_values` returned by the model at a previous stage of decoding, when `use_cache=True` or `config.use_cache=True`.  Only [Cache](/docs/transformers/v5.20.0/en/internal/generation_utils#transformers.Cache) instance is allowed as input, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache). If no `past_key_values` are passed, [DynamicCache](/docs/transformers/v5.20.0/en/internal/generation_utils#transformers.DynamicCache) will be initialized by default.  The model will output the same cache format that is fed as input.  If `past_key_values` are used, the user is expected to input only unprocessed `input_ids` (those that don't have their past key value states given to this model) of shape `(batch_size, unprocessed_length)` instead of all `input_ids` of shape `(batch_size, sequence_length)`.

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.

start_positions (`torch.LongTensor` of shape `(batch_size,)`, *optional*) : Labels for position (index) of the start of the labelled span for computing the token classification loss. Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the sequence are not taken into account for computing the loss.

end_positions (`torch.LongTensor` of shape `(batch_size,)`, *optional*) : Labels for position (index) of the end of the labelled span for computing the token classification loss. Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the sequence are not taken into account for computing the loss.

**Returns:** `QuestionAnsweringModelOutput`

The `GenericForQuestionAnswering` 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.