# coding=utf-8
# Copyright 2023 The Bigcode team and HuggingFace Inc. team.
# 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 GPTBigCode model."""

import math
from typing import Callable, Optional, Union

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

from ...activations import ACT2FN
from ...cache_utils import Cache, EncoderDecoderCache
from ...generation import GenerationMixin
from ...masking_utils import create_causal_mask
from ...modeling_flash_attention_utils import is_flash_attn_available
from ...modeling_outputs import (
    BaseModelOutputWithPastAndCrossAttentions,
    CausalLMOutputWithCrossAttentions,
    SequenceClassifierOutputWithPast,
    TokenClassifierOutput,
)
from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel
from ...utils import (
    auto_docstring,
    can_return_tuple,
    logging,
)
from .configuration_gpt_bigcode import GPTBigCodeConfig


if is_flash_attn_available():
    pass


logger = logging.get_logger(__name__)


# Fused kernels
# Use separate functions for each case because conditionals prevent kernel fusion.
# TODO: Could have better fused kernels depending on scaling, dropout and head mask.
#  Is it doable without writing 32 functions?
@torch.jit.script
def upcast_masked_softmax(
    x: torch.Tensor, mask: torch.Tensor, mask_value: torch.Tensor, scale: float, softmax_dtype: torch.dtype
):
    input_dtype = x.dtype
    x = x.to(softmax_dtype) * scale
    x = torch.where(mask, x, mask_value)
    x = torch.nn.functional.softmax(x, dim=-1).to(input_dtype)
    return x


@torch.jit.script
def upcast_softmax(x: torch.Tensor, scale: float, softmax_dtype: torch.dtype):
    input_dtype = x.dtype
    x = x.to(softmax_dtype) * scale
    x = torch.nn.functional.softmax(x, dim=-1).to(input_dtype)
    return x


@torch.jit.script
def masked_softmax(x: torch.Tensor, mask: torch.Tensor, mask_value: torch.Tensor):
    x = torch.where(mask, x, mask_value)
    x = torch.nn.functional.softmax(x, dim=-1)
    return x


def repeat_kv(hidden_states: torch.Tensor, n_rep: int) -> torch.Tensor:
    """
    This is the equivalent of torch.repeat_interleave(x, dim=1, repeats=n_rep). The hidden states go from (batch,
    num_key_value_heads, seqlen, head_dim) to (batch, num_attention_heads, seqlen, head_dim)
    """
    batch, num_key_value_heads, slen, head_dim = hidden_states.shape
    if n_rep == 1:
        return hidden_states
    hidden_states = hidden_states[:, :, None, :, :].expand(batch, num_key_value_heads, n_rep, slen, head_dim)
    return hidden_states.reshape(batch, num_key_value_heads * n_rep, slen, head_dim)


def eager_attention_forward(
    module: nn.Module,
    query: torch.Tensor,
    key: torch.Tensor,
    value: torch.Tensor,
    attention_mask: Optional[torch.Tensor],
    scaling: float,
    dropout: float = 0.0,
    head_mask: Optional[torch.Tensor] = None,
    **kwargs,
):
    key_states = repeat_kv(key, module.num_key_value_groups)
    value_states = repeat_kv(value, module.num_key_value_groups)

    attn_weights = torch.matmul(query, key_states.transpose(2, 3)) * scaling
    if attention_mask is not None:
        causal_mask = attention_mask[:, :, :, : key_states.shape[-2]]
        attn_weights = attn_weights + causal_mask

    attn_weights = nn.functional.softmax(attn_weights, dim=-1, dtype=torch.float32).to(query.dtype)
    attn_weights = nn.functional.dropout(attn_weights, p=dropout, training=module.training)

    if head_mask is not None:
        attn_weights = attn_weights * head_mask.view(1, -1, 1, 1)

    attn_output = torch.matmul(attn_weights, value_states)
    attn_output = attn_output.transpose(1, 2).contiguous()

    return attn_output, attn_weights


class GPTBigCodeAttention(nn.Module):
    def __init__(self, config, is_cross_attention=False, layer_idx=None):
        super().__init__()
        self.config = config

        self.mask_value = None
        self.multi_query = config.multi_query
        self.embed_dim = config.hidden_size
        self.num_heads = config.num_attention_heads
        self.head_dim = self.embed_dim // self.num_heads
        self.kv_heads = 1 if self.multi_query else self.num_heads
        self.kv_dim = self.kv_heads * self.head_dim
        self.num_key_value_groups = self.num_heads // self.kv_heads
        self.split_size = self.embed_dim
        self.is_causal = True

        if self.head_dim * self.num_heads != self.embed_dim:
            raise ValueError(
                f"`embed_dim` must be divisible by num_heads (got `embed_dim`: {self.embed_dim} and `num_heads`:"
                f" {self.num_heads})."
            )

        self.scale_attn_weights = config.scale_attn_weights
        self.scaling = self.head_dim**0.5 if config.scale_attn_weights else 1.0
        self.is_cross_attention = is_cross_attention

        self.layer_idx = layer_idx
        self.attention_softmax_in_fp32 = config.attention_softmax_in_fp32
        self.scale_attention_softmax_in_fp32 = (
            config.scale_attention_softmax_in_fp32 and config.attention_softmax_in_fp32
        )
        self.attn_pdrop = config.attn_pdrop

        if self.is_cross_attention:
            if self.multi_query:
                raise NotImplementedError("Multi-Query Attention not supported for cross_attention")

            self.c_attn = nn.Linear(self.embed_dim, 2 * self.embed_dim)
            self.q_attn = nn.Linear(self.embed_dim, self.embed_dim)
        else:
            self.c_attn = nn.Linear(self.embed_dim, self.embed_dim + 2 * self.kv_dim)

        self.c_proj = nn.Linear(self.embed_dim, self.embed_dim)

        self.attn_dropout = config.attn_pdrop
        self.resid_dropout = nn.Dropout(config.resid_pdrop)

    def forward(
        self,
        hidden_states: torch.Tensor,
        layer_past: Optional[Cache] = None,
        attention_mask: Optional[torch.Tensor] = None,
        head_mask: Optional[torch.Tensor] = None,
        encoder_hidden_states: Optional[torch.Tensor] = None,
        encoder_attention_mask: Optional[torch.Tensor] = None,
        use_cache: Optional[bool] = False,
        output_attentions: Optional[bool] = False,
        cache_position: Optional[torch.Tensor] = None,
        **kwargs,
    ) -> Union[
        tuple[torch.Tensor, Optional[torch.Tensor]],
        tuple[torch.Tensor, Optional[torch.Tensor], tuple[torch.Tensor, ...]],
    ]:
        input_shape = hidden_states.shape[:-1]

        if layer_past is not None:
            if isinstance(layer_past, EncoderDecoderCache):
                is_updated = layer_past.is_updated.get(self.layer_idx)
                if self.is_cross_attention:
                    # after the first generated id, we can subsequently re-use all key/value_states from cache
                    curr_past_key_value = layer_past.cross_attention_cache
                else:
                    curr_past_key_value = layer_past.self_attention_cache
            else:
                curr_past_key_value = layer_past

        if self.is_cross_attention:
            if not hasattr(self, "q_attn") or not self.is_cross_attention:
                raise ValueError(
                    "If class is used as cross attention, the weights `q_attn` have to be defined. "
                    "Please make sure to instantiate class with `GPTBigCodeAttention(..., is_cross_attention=True)`."
                )
            if layer_past is not None and is_updated:
                # reuse k,v, cross_attentions
                key = curr_past_key_value.layers[self.layer_idx].keys
                value = curr_past_key_value.layers[self.layer_idx].values
            else:
                query = self.q_attn(hidden_states).view(*input_shape, -1, self.head_dim).transpose(1, 2)
                key, value = self.c_attn(encoder_hidden_states).split((self.head_dim, self.head_dim), dim=-1)
        else:
            if self.multi_query:
                query, key, value = (
                    self.c_attn(hidden_states).unsqueeze(1).split((self.embed_dim, self.kv_dim, self.kv_dim), dim=3)
                )
                query = query.view(*input_shape, -1, self.head_dim).transpose(1, 2)
            else:
                query, key, value = (
                    self.c_attn(hidden_states)
                    .view(*hidden_states.shape[:2], self.num_heads, 3 * self.head_dim)
                    .transpose(1, 2)
                    .split(3 * [self.head_dim], dim=3)
                )

        if layer_past is not None:
            # save all key/value_states to cache to be re-used for fast auto-regressive generation
            cache_position = cache_position if not self.is_cross_attention else None
            key, value = curr_past_key_value.update(key, value, self.layer_idx, {"cache_position": cache_position})
            # set flag that curr layer for cross-attn is already updated so we can re-use in subsequent calls
            if self.is_cross_attention:
                layer_past.is_updated[self.layer_idx] = True

        attention_interface: Callable = eager_attention_forward
        if self.config._attn_implementation != "eager":
            attention_interface = ALL_ATTENTION_FUNCTIONS[self.config._attn_implementation]

        attn_output, attn_weights = attention_interface(
            self,
            query,
            key,
            value,
            attention_mask,
            dropout=0.0 if not self.training else self.attn_dropout,
            scaling=self.scaling,
            head_mask=head_mask,
            **kwargs,
        )

        attn_output = attn_output.reshape(*input_shape, -1).contiguous()
        attn_output = self.c_proj(attn_output)
        attn_output = self.resid_dropout(attn_output)
        return attn_output, attn_weights


class GPTBigCodeMLP(nn.Module):
    def __init__(self, intermediate_size, config):
        super().__init__()
        embed_dim = config.hidden_size
        self.c_fc = nn.Linear(embed_dim, intermediate_size)
        self.c_proj = nn.Linear(intermediate_size, embed_dim)
        self.act = ACT2FN[config.activation_function]
        self.dropout = nn.Dropout(config.resid_pdrop)

    # Copied from transformers.models.gpt2.modeling_gpt2.GPT2MLP.forward
    def forward(self, hidden_states: Optional[tuple[torch.FloatTensor]]) -> torch.FloatTensor:
        hidden_states = self.c_fc(hidden_states)
        hidden_states = self.act(hidden_states)
        hidden_states = self.c_proj(hidden_states)
        hidden_states = self.dropout(hidden_states)
        return hidden_states


class GPTBigCodeBlock(nn.Module):
    def __init__(self, config, layer_idx=None):
        super().__init__()
        hidden_size = config.hidden_size
        self.inner_dim = config.n_inner if config.n_inner is not None else 4 * hidden_size

        self.ln_1 = nn.LayerNorm(hidden_size, eps=config.layer_norm_epsilon)

        self.attn = GPTBigCodeAttention(config, layer_idx=layer_idx)

        self.ln_2 = nn.LayerNorm(hidden_size, eps=config.layer_norm_epsilon)

        if config.add_cross_attention:
            if config.multi_query:
                raise NotImplementedError("Cross-attention not implemented for MQA")

            self.crossattention = GPTBigCodeAttention(config, is_cross_attention=True, layer_idx=layer_idx)

            self.ln_cross_attn = nn.LayerNorm(hidden_size, eps=config.layer_norm_epsilon)

        self.mlp = GPTBigCodeMLP(self.inner_dim, config)

    def forward(
        self,
        hidden_states: Optional[tuple[torch.Tensor]],
        layer_past: Optional[Cache] = None,
        attention_mask: Optional[torch.Tensor] = None,
        head_mask: Optional[torch.Tensor] = None,
        encoder_hidden_states: Optional[torch.Tensor] = None,
        encoder_attention_mask: Optional[torch.Tensor] = None,
        use_cache: Optional[bool] = False,
        output_attentions: Optional[bool] = False,
        cache_position: Optional[torch.Tensor] = None,
        **kwargs,
    ) -> Union[
        tuple[torch.Tensor], tuple[torch.Tensor, torch.Tensor], tuple[torch.Tensor, torch.Tensor, torch.Tensor]
    ]:
        residual = hidden_states
        hidden_states = self.ln_1(hidden_states)
        attn_outputs = self.attn(
            hidden_states,
            layer_past=layer_past,
            attention_mask=attention_mask,
            head_mask=head_mask,
            use_cache=use_cache,
            output_attentions=output_attentions,
            cache_position=cache_position,
            **kwargs,
        )
        attn_output = attn_outputs[0]  # output_attn: a, present, (attentions)
        outputs = attn_outputs[1:]
        # residual connection
        hidden_states = attn_output + residual

        if encoder_hidden_states is not None:
            # add one self-attention block for cross-attention
            if not hasattr(self, "crossattention"):
                raise ValueError(
                    f"If `encoder_hidden_states` are passed, {self} has to be instantiated with "
                    "cross-attention layers by setting `config.add_cross_attention=True`"
                )
            residual = hidden_states
            hidden_states = self.ln_cross_attn(hidden_states)
            cross_attn_outputs = self.crossattention(
                hidden_states,
                attention_mask=attention_mask,
                head_mask=head_mask,
                encoder_hidden_states=encoder_hidden_states,
                encoder_attention_mask=encoder_attention_mask,
                output_attentions=output_attentions,
                cache_position=cache_position,
                **kwargs,
            )
            attn_output = cross_attn_outputs[0]
            # residual connection
            hidden_states = residual + attn_output
            outputs = outputs + cross_attn_outputs[1:]  # add cross attentions if we output attention weights

        residual = hidden_states
        hidden_states = self.ln_2(hidden_states)
        feed_forward_hidden_states = self.mlp(hidden_states)
        hidden_states = residual + feed_forward_hidden_states
        return (hidden_states,) + outputs


@auto_docstring
class GPTBigCodePreTrainedModel(PreTrainedModel):
    config: GPTBigCodeConfig
    base_model_prefix = "transformer"
    supports_gradient_checkpointing = True
    _no_split_modules = ["GPTBigCodeBlock"]
    _skip_keys_device_placement = "past_key_values"
    _supports_flash_attn = True
    _supports_sdpa = True

    def __init__(self, *inputs, **kwargs):
        super().__init__(*inputs, **kwargs)

    def _init_weights(self, module):
        """Initialize the weights."""
        if isinstance(module, (GPTBigCodeMLP, GPTBigCodeAttention)):
            # Reinitialize selected weights subject to the OpenAI GPT-2 Paper Scheme:
            #   > A modified initialization which accounts for the accumulation on the residual path with model depth. Scale
            #   > the weights of residual layers at initialization by a factor of 1/√N where N is the # of residual layers.
            #   >   -- GPT-2 :: https://openai.com/blog/better-language-models/
            #
            # Reference (Megatron-LM): https://github.com/NVIDIA/Megatron-LM/blob/main/megatron/model/gpt_model.py
            module.c_proj.weight.data.normal_(
                mean=0.0, std=(self.config.initializer_range / math.sqrt(2 * self.config.n_layer))
            )
            module.c_proj._is_hf_initialized = True
        elif isinstance(module, nn.Linear):
            # Slightly different from the TF version which uses truncated_normal for initialization
            # cf https://github.com/pytorch/pytorch/pull/5617
            module.weight.data.normal_(mean=0.0, std=self.config.initializer_range)
            if module.bias is not None:
                module.bias.data.zero_()
        elif isinstance(module, nn.Embedding):
            module.weight.data.normal_(mean=0.0, std=self.config.initializer_range)
            if module.padding_idx is not None:
                module.weight.data[module.padding_idx].zero_()
        elif isinstance(module, nn.LayerNorm):
            module.bias.data.zero_()
            module.weight.data.fill_(1.0)


@auto_docstring
class GPTBigCodeModel(GPTBigCodePreTrainedModel):
    def __init__(self, config):
        super().__init__(config)
        self.multi_query = config.multi_query
        self.embed_dim = config.hidden_size

        self.wte = nn.Embedding(config.vocab_size, self.embed_dim)
        self.wpe = nn.Embedding(config.max_position_embeddings, self.embed_dim)

        self.drop = nn.Dropout(config.embd_pdrop)
        self.h = nn.ModuleList([GPTBigCodeBlock(config, layer_idx=i) for i in range(config.num_hidden_layers)])
        self.ln_f = nn.LayerNorm(self.embed_dim, eps=config.layer_norm_epsilon)

        max_positions = config.max_position_embeddings
        self.register_buffer(
            "bias", torch.tril(torch.ones((max_positions, max_positions), dtype=torch.bool)), persistent=False
        )

        self.gradient_checkpointing = False

        self._use_sdpa = config._attn_implementation == "sdpa"
        self._use_flash_attention_2 = config._attn_implementation == "flash_attention_2"

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

    def get_input_embeddings(self):
        return self.wte

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

    @can_return_tuple
    @auto_docstring
    def forward(
        self,
        input_ids: Optional[torch.Tensor] = None,
        past_key_values: Optional[list[torch.Tensor]] = None,
        attention_mask: Optional[torch.Tensor] = None,
        token_type_ids: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.Tensor] = None,
        head_mask: Optional[torch.Tensor] = None,
        inputs_embeds: Optional[torch.Tensor] = None,
        encoder_hidden_states: Optional[torch.Tensor] = None,
        encoder_attention_mask: Optional[torch.Tensor] = None,
        use_cache: Optional[bool] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
        cache_position: Optional[torch.Tensor] = None,
        **kwargs,
    ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]:
        r"""
        input_ids (`torch.Tensor` of shape `(batch_size, input_ids_length)`):
            `input_ids_length` = `sequence_length` if `past_key_values` is `None` else
            `past_key_values.get_seq_length()` (`sequence_length` of input past key value states). Indices of input
            sequence tokens in the vocabulary.

            If `past_key_values` is used, only `input_ids` that do not have their past calculated should be passed as
            `input_ids`.

            Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and
            [`PreTrainedTokenizer.__call__`] for details.

            [What are input IDs?](../glossary#input-ids)
        """
        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
        )
        use_cache = use_cache if use_cache is not None else self.config.use_cache
        return_dict = return_dict if return_dict is not None else self.config.use_return_dict

        if (input_ids is None) ^ (inputs_embeds is not None):
            raise ValueError("You must specify exactly one of input_ids or inputs_embeds")
        elif input_ids is not None:
            input_shape = input_ids.size()
            input_ids = input_ids.view(-1, input_shape[-1])
            batch_size = input_ids.shape[0]
        elif inputs_embeds is not None:
            input_shape = inputs_embeds.size()[:-1]
            batch_size = inputs_embeds.shape[0]
        else:
            raise ValueError("You have to specify either input_ids or inputs_embeds")

        if batch_size <= 0:
            raise ValueError("batch_size has to be defined and > 0")

        return_legacy_cache = False
        if use_cache and not isinstance(past_key_values, Cache):
            logger.warning_once(
                "Passing a tuple of `past_key_values` is deprecated and will be removed in Transformers v4.58.0. "
                "You should pass an instance of `EncoderDecoderCache` instead, e.g. "
                "`past_key_values=EncoderDecoderCache.from_legacy_cache(past_key_values)`."
            )
            return_legacy_cache = True
            past_key_values = EncoderDecoderCache.from_legacy_cache(past_key_values)

        if inputs_embeds is None:
            inputs_embeds = self.wte(input_ids)

        if cache_position is None:
            past_seen_tokens = past_key_values.get_seq_length() if past_key_values is not None else 0
            cache_position = torch.arange(
                past_seen_tokens, past_seen_tokens + inputs_embeds.shape[1], device=inputs_embeds.device
            )

        if position_ids is None:
            position_ids = cache_position.unsqueeze(0)

        causal_mask = create_causal_mask(
            config=self.config,
            input_embeds=inputs_embeds,
            attention_mask=attention_mask,
            cache_position=cache_position,
            position_ids=position_ids,
            past_key_values=past_key_values,
        )

        if self._use_flash_attention_2:
            encoder_attention_mask = (
                encoder_attention_mask.bool()
                if (encoder_attention_mask is not None and 0 in encoder_attention_mask)
                else None
            )
        else:
            # If a 2D or 3D attention mask is provided for the cross-attention
            # we need to make broadcastable to [batch_size, num_heads, seq_length, seq_length]
            if (
                self.config.add_cross_attention
                and encoder_hidden_states is not None
                and encoder_attention_mask is not None
            ):
                if encoder_attention_mask.dim() == 2:
                    encoder_attention_mask.unsqueeze(1)
                assert encoder_attention_mask.dim() == 3
                encoder_attention_mask = encoder_attention_mask.bool().unsqueeze(2 if self.multi_query else 1)
            else:
                encoder_attention_mask = None

        # Prepare head mask if needed
        # 1.0 in head_mask indicate we keep the head
        # attention_probs has shape bsz x n_heads x N x N
        # head_mask has shape n_layer x batch x n_heads x N x N
        head_mask = self.get_head_mask(head_mask, self.config.n_layer)

        position_embeds = self.wpe(position_ids)
        hidden_states = inputs_embeds + position_embeds.to(inputs_embeds.device)

        if token_type_ids is not None:
            token_type_ids = token_type_ids.view(-1, input_shape[-1])
            token_type_embeds = self.wte(token_type_ids)
            hidden_states = hidden_states + token_type_embeds

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

        all_self_attentions = () if output_attentions else None
        all_cross_attentions = () if output_attentions and self.config.add_cross_attention 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,
                past_key_values,
                causal_mask,
                head_mask[i],
                encoder_hidden_states,  # as a positional argument for gradient checkpointing
                encoder_attention_mask=encoder_attention_mask,
                use_cache=use_cache,
                output_attentions=output_attentions,
                cache_position=cache_position,
                **kwargs,
            )

            hidden_states = outputs[0]
            if output_attentions:
                all_self_attentions = all_self_attentions + (outputs[1],)
                if self.config.add_cross_attention:
                    all_cross_attentions = all_cross_attentions + (outputs[2],)

        hidden_states = self.ln_f(hidden_states)

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

        if return_legacy_cache:
            past_key_values = past_key_values.to_legacy_cache()

        return BaseModelOutputWithPastAndCrossAttentions(
            last_hidden_state=hidden_states,
            past_key_values=past_key_values,
            hidden_states=all_hidden_states,
            attentions=all_self_attentions,
            cross_attentions=all_cross_attentions,
        )


@auto_docstring(
    custom_intro="""
    The GPT_BIGCODE Model transformer with a language modeling head on top (linear layer with weights tied to the input
    embeddings).
    """
)
class GPTBigCodeForCausalLM(GPTBigCodePreTrainedModel, GenerationMixin):
    _tied_weights_keys = ["lm_head.weight"]

    def __init__(self, config):
        super().__init__(config)
        self.transformer = GPTBigCodeModel(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: Optional[torch.Tensor] = None,
        past_key_values: Optional[tuple[tuple[torch.Tensor]]] = None,
        attention_mask: Optional[torch.Tensor] = None,
        token_type_ids: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.Tensor] = None,
        head_mask: Optional[torch.Tensor] = None,
        inputs_embeds: Optional[torch.Tensor] = None,
        encoder_hidden_states: Optional[torch.Tensor] = None,
        encoder_attention_mask: Optional[torch.Tensor] = None,
        labels: Optional[torch.Tensor] = None,
        use_cache: Optional[bool] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
        cache_position: Optional[torch.Tensor] = None,
        **kwargs,
    ) -> Union[tuple, CausalLMOutputWithCrossAttentions]:
        r"""
        input_ids (`torch.Tensor` of shape `(batch_size, input_ids_length)`):
            `input_ids_length` = `sequence_length` if `past_key_values` is `None` else
            `past_key_values.get_seq_length()` (`sequence_length` of input past key value states). Indices of input
            sequence tokens in the vocabulary.

            If `past_key_values` is used, only `input_ids` that do not have their past calculated should be passed as
            `input_ids`.

            Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and
            [`PreTrainedTokenizer.__call__`] for details.

            [What are input IDs?](../glossary#input-ids)
        labels (`torch.Tensor` of shape `(batch_size, input_ids_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.use_return_dict

        transformer_outputs = self.transformer(
            input_ids,
            past_key_values=past_key_values,
            attention_mask=attention_mask,
            token_type_ids=token_type_ids,
            position_ids=position_ids,
            head_mask=head_mask,
            inputs_embeds=inputs_embeds,
            encoder_hidden_states=encoder_hidden_states,
            encoder_attention_mask=encoder_attention_mask,
            use_cache=use_cache,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
            cache_position=cache_position,
        )
        hidden_states = transformer_outputs[0]

        lm_logits = self.lm_head(hidden_states)

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

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

        return CausalLMOutputWithCrossAttentions(
            loss=loss,
            logits=lm_logits,
            past_key_values=transformer_outputs.past_key_values,
            hidden_states=transformer_outputs.hidden_states,
            attentions=transformer_outputs.attentions,
            cross_attentions=transformer_outputs.cross_attentions,
        )


@auto_docstring(
    custom_intro="""
    The GPTBigCode Model transformer with a sequence classification head on top (linear layer).

    [`GPTBigCodeForSequenceClassification`] uses the last token in order to do the classification, as other causal
    models (e.g. GPT-1) 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 GPTBigCodeForSequenceClassification(GPTBigCodePreTrainedModel):
    def __init__(self, config):
        super().__init__(config)
        self.num_labels = config.num_labels
        self.transformer = GPTBigCodeModel(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: Optional[torch.Tensor] = None,
        past_key_values: Optional[tuple[tuple[torch.Tensor]]] = None,
        attention_mask: Optional[torch.Tensor] = None,
        token_type_ids: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.Tensor] = None,
        head_mask: Optional[torch.Tensor] = None,
        inputs_embeds: Optional[torch.Tensor] = None,
        labels: Optional[torch.Tensor] = None,
        use_cache: Optional[bool] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
        **kwargs,
    ) -> Union[tuple, SequenceClassifierOutputWithPast]:
        r"""
        input_ids (`torch.Tensor` of shape `(batch_size, input_ids_length)`):
            `input_ids_length` = `sequence_length` if `past_key_values` is `None` else
            `past_key_values.get_seq_length()` (`sequence_length` of input past key value states). Indices of input
            sequence tokens in the vocabulary.

            If `past_key_values` is used, only `input_ids` that do not have their past calculated should be passed as
            `input_ids`.

            Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and
            [`PreTrainedTokenizer.__call__`] for details.

            [What are input IDs?](../glossary#input-ids)
        labels (`torch.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).
        """
        return_dict = return_dict if return_dict is not None else self.config.use_return_dict

        transformer_outputs = self.transformer(
            input_ids,
            past_key_values=past_key_values,
            attention_mask=attention_mask,
            token_type_ids=token_type_ids,
            position_ids=position_ids,
            head_mask=head_mask,
            inputs_embeds=inputs_embeds,
            use_cache=use_cache,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
            **kwargs,
        )
        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]

        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:
            labels = labels.to(logits.device)

            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 SequenceClassifierOutputWithPast(
            loss=loss,
            logits=pooled_logits,
            past_key_values=transformer_outputs.past_key_values,
            hidden_states=transformer_outputs.hidden_states,
            attentions=transformer_outputs.attentions,
        )


@auto_docstring
class GPTBigCodeForTokenClassification(GPTBigCodePreTrainedModel):
    def __init__(self, config):
        super().__init__(config)
        self.num_labels = config.num_labels

        self.transformer = GPTBigCodeModel(config)
        if hasattr(config, "classifier_dropout") and config.classifier_dropout is not None:
            classifier_dropout = config.classifier_dropout
        elif hasattr(config, "hidden_dropout") and config.hidden_dropout is not None:
            classifier_dropout = config.hidden_dropout
        else:
            classifier_dropout = 0.1
        self.dropout = nn.Dropout(classifier_dropout)
        self.classifier = nn.Linear(config.hidden_size, config.num_labels)

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

    @auto_docstring
    def forward(
        self,
        input_ids: Optional[torch.Tensor] = None,
        past_key_values: Optional[tuple[tuple[torch.Tensor]]] = None,
        attention_mask: Optional[torch.Tensor] = None,
        token_type_ids: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.Tensor] = None,
        head_mask: Optional[torch.Tensor] = None,
        inputs_embeds: Optional[torch.Tensor] = None,
        labels: Optional[torch.Tensor] = None,
        use_cache: Optional[bool] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
    ) -> Union[tuple, TokenClassifierOutput]:
        r"""
        input_ids (`torch.Tensor` of shape `(batch_size, input_ids_length)`):
            `input_ids_length` = `sequence_length` if `past_key_values` is `None` else
            `past_key_values.get_seq_length()` (`sequence_length` of input past key value states). Indices of input
            sequence tokens in the vocabulary.

            If `past_key_values` is used, only `input_ids` that do not have their past calculated should be passed as
            `input_ids`.

            Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and
            [`PreTrainedTokenizer.__call__`] for details.

            [What are input IDs?](../glossary#input-ids)
        labels (`torch.Tensor` of shape `(batch_size, sequence_length)`, *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.use_return_dict

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

        hidden_states = transformer_outputs[0]
        hidden_states = self.dropout(hidden_states)
        logits = self.classifier(hidden_states)

        loss = None
        if labels is not None:
            loss_fct = CrossEntropyLoss()
            loss = loss_fct(logits.view(-1, self.num_labels), labels.view(-1).to(logits.device))

        if not return_dict:
            output = (logits,) + transformer_outputs[2:]
            return ((loss,) + output) if loss is not None else output

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


__all__ = [
    "GPTBigCodeForSequenceClassification",
    "GPTBigCodeForTokenClassification",
    "GPTBigCodeForCausalLM",
    "GPTBigCodeModel",
    "GPTBigCodePreTrainedModel",
]