image-match/python/py/Lib/site-packages/transformers/models/openai/modeling_openai.py

# Copyright 2018 The OpenAI Team Authors and HuggingFace Inc. team.
# Copyright (c) 2018, NVIDIA CORPORATION.  All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""PyTorch OpenAI GPT model."""

import math
from collections.abc import Callable
from dataclasses import dataclass
from typing import Any

import torch
from torch import nn
from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss

from ... import initialization as init
from ...activations import gelu_new, get_activation, silu
from ...generation import GenerationMixin
from ...modeling_outputs import BaseModelOutput, CausalLMOutput, SequenceClassifierOutput
from ...modeling_utils import PreTrainedModel
from ...pytorch_utils import Conv1D
from ...utils import (
    ModelOutput,
    auto_docstring,
    logging,
)
from .configuration_openai import OpenAIGPTConfig


logger = logging.get_logger(__name__)


ACT_FNS = {"relu": nn.ReLU(), "silu": silu, "gelu": gelu_new, "swish": silu}


class Attention(nn.Module):
    def __init__(self, nx, n_positions, config, scale=False):
        super().__init__()
        self.n_positions = n_positions
        n_state = nx  # in Attention: n_state=768 (nx=n_embd)
        if n_state % config.n_head != 0:
            raise ValueError(f"Attention n_state shape: {n_state} must be divisible by config.n_head {config.n_head}")
        self.register_buffer(
            "bias",
            torch.tril(torch.ones(n_positions, n_positions)).view(1, 1, n_positions, n_positions),
            persistent=False,
        )
        self.n_head = config.n_head
        self.split_size = n_state
        self.scale = scale

        self.c_attn = Conv1D(n_state * 3, nx)
        self.c_proj = Conv1D(n_state, nx)
        self.attn_dropout = nn.Dropout(config.attn_pdrop)
        self.resid_dropout = nn.Dropout(config.resid_pdrop)

    def _attn(self, q, k, v, attention_mask=None, output_attentions=False):
        w = torch.matmul(q, k)
        if self.scale:
            w = w / math.sqrt(v.size(-1))
        # XD: self.b may be larger than w, so we need to crop it
        b = self.bias[:, :, : w.size(-2), : w.size(-1)]
        w = w * b + -1e4 * (1 - b)

        if attention_mask is not None:
            # Apply the attention mask
            w = w + attention_mask

        w = nn.functional.softmax(w, dim=-1)
        w = self.attn_dropout(w)

        outputs = [torch.matmul(w, v)]
        if output_attentions:
            outputs.append(w)
        return outputs

    def merge_heads(self, x):
        x = x.permute(0, 2, 1, 3).contiguous()
        new_x_shape = x.size()[:-2] + (x.size(-2) * x.size(-1),)
        return x.view(*new_x_shape)

    def split_heads(self, x, k=False):
        new_x_shape = x.size()[:-1] + (self.n_head, x.size(-1) // self.n_head)
        x = x.view(*new_x_shape)
        if k:
            return x.permute(0, 2, 3, 1)
        else:
            return x.permute(0, 2, 1, 3)

    def forward(self, x, attention_mask=None, output_attentions=False):
        x = self.c_attn(x)
        query, key, value = x.split(self.split_size, dim=2)
        query = self.split_heads(query)
        key = self.split_heads(key, k=True)
        value = self.split_heads(value)

        attn_outputs = self._attn(query, key, value, attention_mask, output_attentions)
        a = attn_outputs[0]

        a = self.merge_heads(a)
        a = self.c_proj(a)
        a = self.resid_dropout(a)

        outputs = [a] + attn_outputs[1:]
        return outputs  # a, (attentions)


class MLP(nn.Module):
    def __init__(self, n_state, config):  # in MLP: n_state=3072 (4 * n_embd)
        super().__init__()
        nx = config.n_embd
        self.c_fc = Conv1D(n_state, nx)
        self.c_proj = Conv1D(nx, n_state)
        self.act = ACT_FNS[config.afn]
        self.dropout = nn.Dropout(config.resid_pdrop)

    def forward(self, x):
        h = self.act(self.c_fc(x))
        h2 = self.c_proj(h)
        return self.dropout(h2)


class Block(nn.Module):
    def __init__(self, n_positions, config, scale=False):
        super().__init__()
        nx = config.n_embd
        self.attn = Attention(nx, n_positions, config, scale)
        self.ln_1 = nn.LayerNorm(nx, eps=config.layer_norm_epsilon)
        self.mlp = MLP(4 * nx, config)
        self.ln_2 = nn.LayerNorm(nx, eps=config.layer_norm_epsilon)

    def forward(self, x, attention_mask=None, output_attentions=False):
        attn_outputs = self.attn(
            x,
            attention_mask=attention_mask,
            output_attentions=output_attentions,
        )
        a = attn_outputs[0]

        n = self.ln_1(x + a)
        m = self.mlp(n)
        h = self.ln_2(n + m)

        outputs = [h] + attn_outputs[1:]
        return outputs


# Copied from transformers.models.xlm.modeling_xlm.XLMSequenceSummary with XLM->OpenAIGPT
class OpenAIGPTSequenceSummary(nn.Module):
    r"""
    Compute a single vector summary of a sequence hidden states.

    Args:
        config ([`OpenAIGPTConfig`]):
            The config used by the model. Relevant arguments in the config class of the model are (refer to the actual
            config class of your model for the default values it uses):

            - **summary_type** (`str`) -- The method to use to make this summary. Accepted values are:

                - `"last"` -- Take the last token hidden state (like XLNet)
                - `"first"` -- Take the first token hidden state (like Bert)
                - `"mean"` -- Take the mean of all tokens hidden states
                - `"cls_index"` -- Supply a Tensor of classification token position (GPT/GPT-2)
                - `"attn"` -- Not implemented now, use multi-head attention

            - **summary_use_proj** (`bool`) -- Add a projection after the vector extraction.
            - **summary_proj_to_labels** (`bool`) -- If `True`, the projection outputs to `config.num_labels` classes
              (otherwise to `config.hidden_size`).
            - **summary_activation** (`Optional[str]`) -- Set to `"tanh"` to add a tanh activation to the output,
              another string or `None` will add no activation.
            - **summary_first_dropout** (`float`) -- Optional dropout probability before the projection and activation.
            - **summary_last_dropout** (`float`)-- Optional dropout probability after the projection and activation.
    """

    def __init__(self, config: OpenAIGPTConfig):
        super().__init__()

        self.summary_type = getattr(config, "summary_type", "last")
        if self.summary_type == "attn":
            # We should use a standard multi-head attention module with absolute positional embedding for that.
            # Cf. https://github.com/zihangdai/xlnet/blob/master/modeling.py#L253-L276
            # We can probably just use the multi-head attention module of PyTorch >=1.1.0
            raise NotImplementedError

        self.summary = nn.Identity()
        if hasattr(config, "summary_use_proj") and config.summary_use_proj:
            if hasattr(config, "summary_proj_to_labels") and config.summary_proj_to_labels and config.num_labels > 0:
                num_classes = config.num_labels
            else:
                num_classes = config.hidden_size
            self.summary = nn.Linear(config.hidden_size, num_classes)

        activation_string = getattr(config, "summary_activation", None)
        self.activation: Callable = get_activation(activation_string) if activation_string else nn.Identity()

        self.first_dropout = nn.Identity()
        if hasattr(config, "summary_first_dropout") and config.summary_first_dropout > 0:
            self.first_dropout = nn.Dropout(config.summary_first_dropout)

        self.last_dropout = nn.Identity()
        if hasattr(config, "summary_last_dropout") and config.summary_last_dropout > 0:
            self.last_dropout = nn.Dropout(config.summary_last_dropout)

    def forward(
        self, hidden_states: torch.FloatTensor, cls_index: torch.LongTensor | None = None
    ) -> torch.FloatTensor:
        """
        Compute a single vector summary of a sequence hidden states.

        Args:
            hidden_states (`torch.FloatTensor` of shape `[batch_size, seq_len, hidden_size]`):
                The hidden states of the last layer.
            cls_index (`torch.LongTensor` of shape `[batch_size]` or `[batch_size, ...]` where ... are optional leading dimensions of `hidden_states`, *optional*):
                Used if `summary_type == "cls_index"` and takes the last token of the sequence as classification token.

        Returns:
            `torch.FloatTensor`: The summary of the sequence hidden states.
        """
        if self.summary_type == "last":
            output = hidden_states[:, -1]
        elif self.summary_type == "first":
            output = hidden_states[:, 0]
        elif self.summary_type == "mean":
            output = hidden_states.mean(dim=1)
        elif self.summary_type == "cls_index":
            if cls_index is None:
                cls_index = torch.full_like(
                    hidden_states[..., :1, :],
                    hidden_states.shape[-2] - 1,
                    dtype=torch.long,
                )
            else:
                cls_index = cls_index.unsqueeze(-1).unsqueeze(-1)
                cls_index = cls_index.expand((-1,) * (cls_index.dim() - 1) + (hidden_states.size(-1),))
            # shape of cls_index: (bsz, XX, 1, hidden_size) where XX are optional leading dim of hidden_states
            output = hidden_states.gather(-2, cls_index).squeeze(-2)  # shape (bsz, XX, hidden_size)
        elif self.summary_type == "attn":
            raise NotImplementedError

        output = self.first_dropout(output)
        output = self.summary(output)
        output = self.activation(output)
        output = self.last_dropout(output)

        return output


@auto_docstring
class OpenAIGPTPreTrainedModel(PreTrainedModel):
    config: OpenAIGPTConfig
    base_model_prefix = "transformer"

    def _init_weights(self, module):
        super()._init_weights(module)
        if isinstance(module, Attention):
            n_positions = module.n_positions
            init.copy_(
                module.bias, torch.tril(torch.ones(n_positions, n_positions)).view(1, 1, n_positions, n_positions)
            )
        elif isinstance(module, OpenAIGPTModel):
            init.copy_(module.position_ids, torch.arange(module.config.n_positions))


@dataclass
@auto_docstring(
    custom_intro="""
    Base class for outputs of models predicting if two sentences are consecutive or not.
    """
)
class OpenAIGPTDoubleHeadsModelOutput(ModelOutput):
    r"""
    loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided):
        Language modeling loss.
    mc_loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `mc_labels` is provided):
        Multiple choice classification loss.
    logits (`torch.FloatTensor` of shape `(batch_size, num_choices, sequence_length, config.vocab_size)`):
        Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).
    mc_logits (`torch.FloatTensor` of shape `(batch_size, num_choices)`):
        Prediction scores of the multiple choice classification head (scores for each choice before SoftMax).
    """

    loss: torch.FloatTensor | None = None
    mc_loss: torch.FloatTensor | None = None
    logits: torch.FloatTensor | None = None
    mc_logits: torch.FloatTensor | None = None
    hidden_states: tuple[torch.FloatTensor] | None = None
    attentions: tuple[torch.FloatTensor] | None = None


@auto_docstring
class OpenAIGPTModel(OpenAIGPTPreTrainedModel):
    def __init__(self, config):
        super().__init__(config)

        self.tokens_embed = nn.Embedding(config.vocab_size, config.n_embd)
        self.positions_embed = nn.Embedding(config.n_positions, config.n_embd)
        self.drop = nn.Dropout(config.embd_pdrop)
        self.h = nn.ModuleList([Block(config.n_positions, config, scale=True) for _ in range(config.n_layer)])

        self.register_buffer("position_ids", torch.arange(config.n_positions), persistent=False)
        # Initialize weights and apply final processing
        self.post_init()

    def get_input_embeddings(self):
        return self.tokens_embed

    def set_input_embeddings(self, new_embeddings):
        self.tokens_embed = new_embeddings

    @auto_docstring
    def forward(
        self,
        input_ids: torch.LongTensor | None = None,
        attention_mask: torch.FloatTensor | None = None,
        token_type_ids: torch.LongTensor | None = None,
        position_ids: torch.LongTensor | None = None,
        inputs_embeds: torch.FloatTensor | None = None,
        output_attentions: bool | None = None,
        output_hidden_states: bool | None = None,
        return_dict: bool | None = None,
        **kwargs,
    ) -> tuple[torch.Tensor] | BaseModelOutput:
        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
        output_hidden_states = (
            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
        )
        return_dict = return_dict if return_dict is not None else self.config.return_dict

        if input_ids is not None and inputs_embeds is not None:
            raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time")
        elif input_ids is not None:
            self.warn_if_padding_and_no_attention_mask(input_ids, attention_mask)
            input_shape = input_ids.size()
            input_ids = input_ids.view(-1, input_shape[-1])
        elif inputs_embeds is not None:
            input_shape = inputs_embeds.size()[:-1]
        else:
            raise ValueError("You have to specify either input_ids or inputs_embeds")

        if position_ids is None:
            # Code is different from when we had a single embedding matrix  from position and token embeddings
            position_ids = self.position_ids[None, : input_shape[-1]]

        # Attention mask.
        if attention_mask is not None:
            # We create a 3D attention mask from a 2D tensor mask.
            # Sizes are [batch_size, 1, 1, to_seq_length]
            # So we can broadcast to [batch_size, num_heads, from_seq_length, to_seq_length]
            # this attention mask is more simple than the triangular masking of causal attention
            # used in OpenAI GPT, we just need to prepare the broadcast dimension here.
            attention_mask = attention_mask.unsqueeze(1).unsqueeze(2)

            # Since attention_mask is 1.0 for positions we want to attend and 0.0 for
            # masked positions, this operation will create a tensor which is 0.0 for
            # positions we want to attend and the dtype's smallest value for masked positions.
            # Since we are adding it to the raw scores before the softmax, this is
            # effectively the same as removing these entirely.
            attention_mask = attention_mask.to(dtype=next(self.parameters()).dtype)  # fp16 compatibility
            attention_mask = (1.0 - attention_mask) * torch.finfo(self.dtype).min

        if inputs_embeds is None:
            inputs_embeds = self.tokens_embed(input_ids)
        position_embeds = self.positions_embed(position_ids)
        if token_type_ids is not None:
            token_type_ids = token_type_ids.view(-1, token_type_ids.size(-1))
            token_type_embeds = self.tokens_embed(token_type_ids)
        else:
            token_type_embeds = 0
        hidden_states = inputs_embeds + position_embeds + token_type_embeds
        hidden_states = self.drop(hidden_states)

        output_shape = input_shape + (hidden_states.size(-1),)

        all_attentions = () if output_attentions else None
        all_hidden_states = () if output_hidden_states else None
        for i, block in enumerate(self.h):
            if output_hidden_states:
                all_hidden_states = all_hidden_states + (hidden_states,)

            outputs = block(hidden_states, attention_mask, output_attentions=output_attentions)
            hidden_states = outputs[0]
            if output_attentions:
                all_attentions = all_attentions + (outputs[1],)

        hidden_states = hidden_states.view(*output_shape)
        # Add last layer
        if output_hidden_states:
            all_hidden_states = all_hidden_states + (hidden_states,)

        if not return_dict:
            return tuple(v for v in [hidden_states, all_hidden_states, all_attentions] if v is not None)

        return BaseModelOutput(
            last_hidden_state=hidden_states,
            hidden_states=all_hidden_states,
            attentions=all_attentions,
        )


@auto_docstring(
    custom_intro="""
    OpenAI GPT Model transformer with a language modeling head on top (linear layer with weights tied to the input
    embeddings).
    """
)
class OpenAIGPTLMHeadModel(OpenAIGPTPreTrainedModel, GenerationMixin):
    _tied_weights_keys = {"lm_head.weight": "transformer.tokens_embed.weight"}

    def __init__(self, config):
        super().__init__(config)
        self.transformer = OpenAIGPTModel(config)
        self.lm_head = nn.Linear(config.n_embd, config.vocab_size, bias=False)

        # Initialize weights and apply final processing
        self.post_init()

    @auto_docstring
    def forward(
        self,
        input_ids: torch.LongTensor | None = None,
        attention_mask: torch.FloatTensor | None = None,
        token_type_ids: torch.LongTensor | None = None,
        position_ids: torch.LongTensor | None = None,
        inputs_embeds: torch.FloatTensor | None = None,
        labels: torch.LongTensor | None = None,
        output_attentions: bool | None = None,
        output_hidden_states: bool | None = None,
        return_dict: bool | None = None,
        logits_to_keep: int | torch.Tensor = 0,
        **kwargs,
    ) -> tuple[torch.Tensor] | CausalLMOutput:
        r"""
        labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
            Labels for language modeling. Note that the labels **are shifted** inside the model, i.e. you can set
            `labels = input_ids` Indices are selected in `[-100, 0, ..., config.vocab_size]` All labels set to `-100`
            are ignored (masked), the loss is only computed for labels in `[0, ..., config.vocab_size]`
        """
        return_dict = return_dict if return_dict is not None else self.config.return_dict

        transformer_outputs = self.transformer(
            input_ids,
            attention_mask=attention_mask,
            token_type_ids=token_type_ids,
            position_ids=position_ids,
            inputs_embeds=inputs_embeds,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )

        hidden_states = transformer_outputs[0]
        # Only compute necessary logits, and do not upcast them to float if we are not computing the loss
        slice_indices = slice(-logits_to_keep, None) if isinstance(logits_to_keep, int) else logits_to_keep
        logits = self.lm_head(hidden_states[:, slice_indices, :])

        loss = None
        if labels is not None:
            loss = self.loss_function(logits=logits, labels=labels, vocab_size=self.config.vocab_size, **kwargs)

        if not return_dict:
            output = (logits,) + transformer_outputs[1:]
            return ((loss,) + output) if loss is not None else output

        return CausalLMOutput(
            loss=loss,
            logits=logits,
            hidden_states=transformer_outputs.hidden_states,
            attentions=transformer_outputs.attentions,
        )

    def prepare_inputs_for_generation(self, input_ids: torch.LongTensor, **kwargs) -> dict[str, Any]:
        # Overwritten -- old model with reduced inputs
        model_inputs = {"input_ids": input_ids}

        # Forward ALL kwargs that are uninitialized (e.g. `use_cache`).
        for key, value in kwargs.items():
            if key not in model_inputs:
                model_inputs[key] = value

        return model_inputs


@auto_docstring(
    custom_intro="""
        OpenAI GPT Model transformer with a language modeling and a multiple-choice classification head on top e.g. for
    RocStories/SWAG tasks. The two heads are two linear layers. The language modeling head has its weights tied to the
    input embeddings, the classification head takes as input the input of a specified classification token index in the
    input sequence).
    """
)
class OpenAIGPTDoubleHeadsModel(OpenAIGPTPreTrainedModel):
    _tied_weights_keys = {"transformer.tokens_embed.weight": "lm_head.weight"}

    def __init__(self, config):
        super().__init__(config)

        config.num_labels = 1
        self.transformer = OpenAIGPTModel(config)
        self.lm_head = nn.Linear(config.n_embd, config.vocab_size, bias=False)
        self.multiple_choice_head = OpenAIGPTSequenceSummary(config)

        # Initialize weights and apply final processing
        self.post_init()

    @auto_docstring
    def forward(
        self,
        input_ids: torch.LongTensor | None = None,
        attention_mask: torch.FloatTensor | None = None,
        token_type_ids: torch.LongTensor | None = None,
        position_ids: torch.LongTensor | None = None,
        inputs_embeds: torch.FloatTensor | None = None,
        mc_token_ids: torch.LongTensor | None = None,
        labels: torch.LongTensor | None = None,
        mc_labels: torch.LongTensor | None = None,
        output_attentions: bool | None = None,
        output_hidden_states: bool | None = None,
        return_dict: bool | None = None,
        **kwargs,
    ) -> tuple[torch.Tensor] | OpenAIGPTDoubleHeadsModelOutput:
        r"""
        mc_token_ids (`torch.LongTensor` of shape `(batch_size, num_choices)`, *optional*, default to index of the last token of the input):
            Index of the classification token in each input sequence. Selected in the range `[0, input_ids.size(-1) -
            1]`.
        labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
            Labels for language modeling. Note that the labels **are shifted** inside the model, i.e. you can set
            `labels = input_ids` Indices are selected in `[-1, 0, ..., config.vocab_size]` All labels set to `-100` are
            ignored (masked), the loss is only computed for labels in `[0, ..., config.vocab_size]`
        mc_labels (`torch.LongTensor` of shape `(batch_size)`, *optional*):
            Labels for computing the multiple choice classification loss. Indices should be in `[0, ..., num_choices]`
            where *num_choices* is the size of the second dimension of the input tensors. (see *input_ids* above)

        Examples:

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

        >>> tokenizer = AutoTokenizer.from_pretrained("openai-community/openai-gpt")
        >>> model = OpenAIGPTDoubleHeadsModel.from_pretrained("openai-community/openai-gpt")
        >>> tokenizer.add_special_tokens(
        ...     {"cls_token": "[CLS]"}
        ... )  # Add a [CLS] to the vocabulary (we should train it also!)
        >>> model.resize_token_embeddings(len(tokenizer))

        >>> choices = ["Hello, my dog is cute [CLS]", "Hello, my cat is cute [CLS]"]
        >>> input_ids = torch.tensor([tokenizer.encode(s) for s in choices]).unsqueeze(0)  # Batch size 1, 2 choices
        >>> mc_token_ids = torch.tensor([input_ids.size(-1) - 1, input_ids.size(-1) - 1]).unsqueeze(0)  # Batch size 1

        >>> outputs = model(input_ids, mc_token_ids=mc_token_ids)
        >>> lm_logits = outputs.logits
        >>> mc_logits = outputs.mc_logits
        ```"""
        return_dict = return_dict if return_dict is not None else self.config.return_dict

        transformer_outputs = self.transformer(
            input_ids,
            attention_mask=attention_mask,
            token_type_ids=token_type_ids,
            position_ids=position_ids,
            inputs_embeds=inputs_embeds,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )
        hidden_states = transformer_outputs[0]

        lm_logits = self.lm_head(hidden_states)
        mc_logits = self.multiple_choice_head(hidden_states, mc_token_ids).squeeze(-1)

        lm_loss, mc_loss = None, None
        if mc_labels is not None:
            loss_fct = CrossEntropyLoss()
            mc_loss = loss_fct(mc_logits.view(-1, mc_logits.size(-1)), mc_labels.view(-1))
        if labels is not None:
            shift_logits = lm_logits[..., :-1, :].contiguous()
            shift_labels = labels[..., 1:].contiguous()
            loss_fct = CrossEntropyLoss()
            lm_loss = loss_fct(shift_logits.view(-1, shift_logits.size(-1)), shift_labels.view(-1))

        if not return_dict:
            output = (lm_logits, mc_logits) + transformer_outputs[1:]
            if mc_loss is not None:
                output = (mc_loss,) + output
            return ((lm_loss,) + output) if lm_loss is not None else output

        return OpenAIGPTDoubleHeadsModelOutput(
            loss=lm_loss,
            mc_loss=mc_loss,
            logits=lm_logits,
            mc_logits=mc_logits,
            hidden_states=transformer_outputs.hidden_states,
            attentions=transformer_outputs.attentions,
        )


@auto_docstring(
    custom_intro="""
    The Original OpenAI GPT Model transformer with a sequence classification head on top (linear layer).
    [`OpenAIGPTForSequenceClassification`] uses the last token in order to do the classification, as other causal
    models (e.g. GPT-2) do. Since it does classification on the last token, it requires to know the position of the
    last token. If a `pad_token_id` is defined in the configuration, it finds the last token that is not a padding
    token in each row. If no `pad_token_id` is defined, it simply takes the last value in each row of the batch. Since
    it cannot guess the padding tokens when `inputs_embeds` are passed instead of `input_ids`, it does the same (take
    the last value in each row of the batch).
    """
)
class OpenAIGPTForSequenceClassification(OpenAIGPTPreTrainedModel):
    def __init__(self, config):
        super().__init__(config)
        self.num_labels = config.num_labels
        self.transformer = OpenAIGPTModel(config)
        self.score = nn.Linear(config.n_embd, self.num_labels, bias=False)

        # Initialize weights and apply final processing
        self.post_init()

    @auto_docstring
    def forward(
        self,
        input_ids: torch.LongTensor | None = None,
        attention_mask: torch.FloatTensor | None = None,
        token_type_ids: torch.LongTensor | None = None,
        position_ids: torch.LongTensor | None = None,
        inputs_embeds: torch.FloatTensor | None = None,
        labels: torch.LongTensor | None = None,
        output_attentions: bool | None = None,
        output_hidden_states: bool | None = None,
        return_dict: bool | None = None,
        **kwargs,
    ) -> tuple[torch.Tensor] | SequenceClassifierOutput:
        r"""
        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).
        """
        return_dict = return_dict if return_dict is not None else self.config.return_dict

        transformer_outputs = self.transformer(
            input_ids,
            attention_mask=attention_mask,
            token_type_ids=token_type_ids,
            position_ids=position_ids,
            inputs_embeds=inputs_embeds,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )

        hidden_states = transformer_outputs[0]
        logits = self.score(hidden_states)

        if input_ids is not None:
            batch_size, sequence_length = input_ids.shape[:2]
        else:
            batch_size, sequence_length = inputs_embeds.shape[:2]

        # Ensure the batch size is > 1 if there is no padding.
        if self.config.pad_token_id is None and batch_size != 1:
            raise ValueError("Cannot handle batch sizes > 1 if no padding token is defined.")
        if self.config.pad_token_id is None:
            last_non_pad_token = -1
        elif input_ids is not None:
            # To handle both left- and right- padding, we take the rightmost token that is not equal to pad_token_id
            non_pad_mask = (input_ids != self.config.pad_token_id).to(logits.device, torch.int32)
            token_indices = torch.arange(input_ids.shape[-1], device=logits.device, dtype=torch.int32)
            last_non_pad_token = (token_indices * non_pad_mask).argmax(-1)
        else:
            last_non_pad_token = -1
            logger.warning_once(
                f"{self.__class__.__name__} will not detect padding tokens in `inputs_embeds`. Results may be "
                "unexpected if using padding tokens in conjunction with `inputs_embeds.`"
            )

        pooled_logits = logits[torch.arange(batch_size, device=logits.device), last_non_pad_token]

        loss = None
        if labels is not None:
            if self.config.problem_type is None:
                if self.num_labels == 1:
                    self.config.problem_type = "regression"
                elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
                    self.config.problem_type = "single_label_classification"
                else:
                    self.config.problem_type = "multi_label_classification"

            if self.config.problem_type == "regression":
                loss_fct = MSELoss()
                if self.num_labels == 1:
                    loss = loss_fct(pooled_logits.squeeze(), labels.squeeze())
                else:
                    loss = loss_fct(pooled_logits, labels)
            elif self.config.problem_type == "single_label_classification":
                loss_fct = CrossEntropyLoss()
                loss = loss_fct(pooled_logits.view(-1, self.num_labels), labels.view(-1))
            elif self.config.problem_type == "multi_label_classification":
                loss_fct = BCEWithLogitsLoss()
                loss = loss_fct(pooled_logits, labels)
        if not return_dict:
            output = (pooled_logits,) + transformer_outputs[1:]
            return ((loss,) + output) if loss is not None else output

        return SequenceClassifierOutput(
            loss=loss,
            logits=pooled_logits,
            hidden_states=transformer_outputs.hidden_states,
            attentions=transformer_outputs.attentions,
        )


__all__ = [
    "OpenAIGPTDoubleHeadsModel",
    "OpenAIGPTForSequenceClassification",
    "OpenAIGPTLMHeadModel",
    "OpenAIGPTModel",
    "OpenAIGPTPreTrainedModel",
]