modeling_xverse.py

# coding=utf-8
# Copyright 2022 EleutherAI and the HuggingFace Inc. team. All rights reserved.
#
# This code is based on EleutherAI's GPT-NeoX library and the GPT-NeoX
# and OPT implementations in this library. It has been modified from its
# original forms to accommodate minor architectural differences compared
# to GPT-NeoX and OPT used by the Meta AI team that trained the model.
#
# 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 XVERSE model."""
import math
from typing import List, Optional, Tuple, Union

import torch
import torch.nn.functional as F
import torch.utils.checkpoint
from torch import nn
from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss

from transformers.activations import ACT2FN
from transformers.modeling_outputs import BaseModelOutputWithPast, CausalLMOutputWithPast
from transformers.modeling_utils import PreTrainedModel
from transformers.utils import add_start_docstrings, add_start_docstrings_to_model_forward, logging, replace_return_docstrings
from transformers.generation.utils import GenerationConfig
from .configuration_xverse import XverseConfig
from xformers import ops as xops


logger = logging.get_logger(__name__)

_CONFIG_FOR_DOC = "XverseConfig"


# Copied from transformers.models.bart.modeling_bart._make_causal_mask
def _make_causal_mask(
    input_ids_shape: torch.Size, dtype: torch.dtype, device: torch.device, past_key_values_length: int = 0
):
    """
    Make causal mask used for bi-directional self-attention.
    """
    bsz, tgt_len = input_ids_shape
    mask = torch.full((tgt_len, tgt_len), torch.tensor(
        torch.finfo(dtype).min, device=device), device=device)
    mask_cond = torch.arange(mask.size(-1), device=device)
    mask.masked_fill_(mask_cond < (mask_cond + 1).view(mask.size(-1), 1), 0)
    mask = mask.to(dtype)

    if past_key_values_length > 0:
        mask = torch.cat([torch.zeros(
            tgt_len, past_key_values_length, dtype=dtype, device=device), mask], dim=-1)
    return mask[None, None, :, :].expand(bsz, 1, tgt_len, tgt_len + past_key_values_length)


# Copied from transformers.models.bart.modeling_bart._expand_mask
def _expand_mask(mask: torch.Tensor, dtype: torch.dtype, tgt_len: Optional[int] = None):
    """
    Expands attention_mask from `[bsz, seq_len]` to `[bsz, 1, tgt_seq_len, src_seq_len]`.
    """
    bsz, src_len = mask.size()
    tgt_len = tgt_len if tgt_len is not None else src_len

    expanded_mask = mask[:, None, None, :].expand(
        bsz, 1, tgt_len, src_len).to(dtype)

    inverted_mask = 1.0 - expanded_mask

    return inverted_mask.masked_fill(inverted_mask.to(torch.bool), torch.finfo(dtype).min)


class XverseRMSNorm(nn.Module):
    def __init__(self, hidden_size, eps=1e-6):
        """
        XverseRMSNorm is equivalent to T5LayerNorm
        """
        super().__init__()
        self.weight = nn.Parameter(torch.ones(hidden_size))
        self.variance_epsilon = eps

    def forward(self, hidden_states):
        input_dtype = hidden_states.dtype
        variance = hidden_states.to(torch.float32).pow(
            2).mean(-1, keepdim=True)
        hidden_states = hidden_states * \
            torch.rsqrt(variance + self.variance_epsilon)

        return (self.weight * hidden_states).to(input_dtype)


class XverseRotaryEmbedding(torch.nn.Module):
    def __init__(self, dim, max_position_embeddings=2048, base=500000, device=None):
        super().__init__()
        self.base = base
        self.dim = dim
        self.max_position_embeddings = max_position_embeddings
        inv_freq = 1.0 / \
            (base ** (torch.arange(0, dim, 2).float().to(device) / dim))
        self.register_buffer("inv_freq", inv_freq)

        # Build here to make `torch.jit.trace` work.
        self.max_seq_len_cached = max_position_embeddings

        t = torch.arange(self.max_seq_len_cached,
                         device=self.inv_freq.device, dtype=self.inv_freq.dtype)
        freqs = torch.einsum("i,j->ij", t, self.inv_freq)
        # Different from paper, but it uses a different permutation in order to obtain the same calculation
        emb = torch.cat((freqs, freqs), dim=-1)
        self.register_buffer("cos_cached", emb.cos()[
                             None, None, :, :], persistent=False)
        self.register_buffer("sin_cached", emb.sin()[
                             None, None, :, :], persistent=False)

    def forward(self, x, seq_len=None):
        # x: [bs, num_attention_heads, seq_len, head_size]
        # This `if` block is unlikely to be run after we build sin/cos in `__init__`. Keep the logic here just in case.
        if seq_len > self.max_seq_len_cached:

            t = torch.arange(seq_len, device=x.device, dtype=torch.float32)
            dim = self.dim
            alpha = (seq_len / (self.max_position_embeddings/2) - 1)
            base = self.base * alpha ** (dim / (dim-2))
            ntk_inv_freq = 1.0 / \
                (base ** (torch.arange(0, dim, 2).float().to(x.device) / dim))

            freqs = torch.einsum("i,j->ij", t, ntk_inv_freq)
            emb = torch.cat((freqs, freqs), dim=-1).to(x.device)
            cos_cached = emb.cos()[None, None, :, :]
            sin_cached = emb.sin()[None, None, :, :]
            return (
                cos_cached[:, :, :seq_len, ...].to(dtype=x.dtype),
                sin_cached[:, :, :seq_len, ...].to(dtype=x.dtype)
            )

        return (
            self.cos_cached[:, :, :seq_len, ...].to(dtype=x.dtype),
            self.sin_cached[:, :, :seq_len, ...].to(dtype=x.dtype),
        )


def rotate_half(x):
    """Rotates half the hidden dims of the input."""
    x1 = x[..., : x.shape[-1] // 2]
    x2 = x[..., x.shape[-1] // 2:]
    return torch.cat((-x2, x1), dim=-1)


def apply_rotary_pos_emb(q, k, cos, sin, position_ids):
    # The first two dimensions of cos and sin are always 1, so we can `squeeze` them.
    cos = cos.squeeze(1).squeeze(0)  # [seq_len, dim]
    sin = sin.squeeze(1).squeeze(0)  # [seq_len, dim]
    cos = cos[position_ids].unsqueeze(1)  # [bs, 1, seq_len, dim]
    sin = sin[position_ids].unsqueeze(1)  # [bs, 1, seq_len, dim]
    q_embed = (q * cos) + (rotate_half(q) * sin)
    k_embed = (k * cos) + (rotate_half(k) * sin)
    return q_embed, k_embed


class XverseMLP(nn.Module):
    def __init__(
        self,
        hidden_size: int,
        intermediate_size: int,
        hidden_act: str,
    ):
        super().__init__()
        self.gate_proj = nn.Linear(hidden_size, intermediate_size, bias=False)
        self.down_proj = nn.Linear(intermediate_size, hidden_size, bias=False)
        self.up_proj = nn.Linear(hidden_size, intermediate_size, bias=False)
        self.act_fn = ACT2FN[hidden_act]

    def forward(self, x):
        return self.down_proj(self.act_fn(self.gate_proj(x)) * self.up_proj(x))


class XverseAttention(nn.Module):
    """Multi-headed attention from 'Attention Is All You Need' paper"""

    def __init__(self, config: XverseConfig):
        super().__init__()
        self.config = config
        self.hidden_size = config.hidden_size
        self.num_heads = config.num_attention_heads
        self.head_dim = self.hidden_size // self.num_heads
        self.max_position_embeddings = config.max_position_embeddings

        if (self.head_dim * self.num_heads) != self.hidden_size:
            raise ValueError(
                f"hidden_size must be divisible by num_heads (got `hidden_size`: {self.hidden_size}"
                f" and `num_heads`: {self.num_heads})."
            )
        self.q_proj = nn.Linear(
            self.hidden_size, self.num_heads * self.head_dim, bias=False)
        self.k_proj = nn.Linear(
            self.hidden_size, self.num_heads * self.head_dim, bias=False)
        self.v_proj = nn.Linear(
            self.hidden_size, self.num_heads * self.head_dim, bias=False)
        self.o_proj = nn.Linear(
            self.num_heads * self.head_dim, self.hidden_size, bias=False)
        self.rotary_emb = XverseRotaryEmbedding(
            self.head_dim, max_position_embeddings=self.max_position_embeddings)

    def _shape(self, tensor: torch.Tensor, seq_len: int, bsz: int):
        return tensor.view(bsz, seq_len, self.num_heads, self.head_dim).transpose(1, 2).contiguous()

    def forward(
        self,
        hidden_states: torch.Tensor,
        attention_mask: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        past_key_value: Optional[Tuple[torch.Tensor]] = None,
        output_attentions: bool = False,
        use_cache: bool = False,
        dropout: Optional[float] = 0.1,
    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:

        bsz, q_len, _ = hidden_states.size()  # [bsz, q_len, hidden_size]

        query_states = self.q_proj(hidden_states).view(
            bsz, q_len, self.num_heads, self.head_dim)
        key_states = self.k_proj(hidden_states).view(
            bsz, q_len, self.num_heads, self.head_dim)
        value_states = self.v_proj(hidden_states).view(
            bsz, q_len, self.num_heads, self.head_dim)
        # [bsz, q_len, nh, hd]

        query_states = query_states.transpose(1, 2)
        key_states = key_states.transpose(1, 2)
        value_states = value_states.transpose(1, 2)
        # [bsz, nh, q_len, hd]

        kv_seq_len = key_states.shape[-2]  # q_len
        n_head = key_states.shape[-3]  # nh

        assert past_key_value is None, "past_key_value is not supported"
        cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
        query_states, key_states = apply_rotary_pos_emb(
            query_states, key_states, cos, sin, position_ids)

        assert not output_attentions, "output_attentions is not supported"
        assert not use_cache, "use_cache is not supported"

        """
        Input tensors must be in format ``[B, M, H, K]``, where B is the batch size, M 
        the sequence length, H the number of heads, and K the embeding size per head
        """
        query_states = query_states.transpose(1, 2)
        key_states = key_states.transpose(1, 2)
        value_states = value_states.transpose(1, 2)

        attn_output = xops.memory_efficient_attention(
            query_states, key_states, value_states, attn_bias=xops.LowerTriangularMask(), p=dropout
        )
        
        attn_output = attn_output.reshape(bsz, q_len, self.hidden_size)
        attn_output = self.o_proj(attn_output)
        return attn_output, None, None


class XverseDecoderLayer(nn.Module):
    def __init__(self, config: XverseConfig):
        super().__init__()
        self.hidden_size = config.hidden_size
        self.self_attn = XverseAttention(config=config)
        self.mlp = XverseMLP(
            hidden_size=self.hidden_size,
            intermediate_size=config.intermediate_size,
            hidden_act=config.hidden_act,
        )
        self.input_layernorm = XverseRMSNorm(
            config.hidden_size, eps=config.rms_norm_eps)
        self.post_attention_layernorm = XverseRMSNorm(
            config.hidden_size, eps=config.rms_norm_eps)

    def forward(
        self,
        hidden_states: torch.Tensor,
        attention_mask: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        past_key_value: Optional[Tuple[torch.Tensor]] = None,
        output_attentions: Optional[bool] = False,
        use_cache: Optional[bool] = False,
    ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
        """
        Args:
            hidden_states (`torch.FloatTensor`): input to the layer of shape `(batch, seq_len, embed_dim)`
            attention_mask (`torch.FloatTensor`, *optional*): attention mask of size
                `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
            output_attentions (`bool`, *optional*):
                Whether or not to return the attentions tensors of all attention layers. See `attentions` under
                returned tensors for more detail.
            use_cache (`bool`, *optional*):
                If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding
                (see `past_key_values`).
            past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states
        """

        residual = hidden_states

        hidden_states = self.input_layernorm(hidden_states)

        # Self Attention
        hidden_states, self_attn_weights, present_key_value = self.self_attn(
            hidden_states=hidden_states,
            attention_mask=attention_mask,
            position_ids=position_ids,
            past_key_value=past_key_value,
            output_attentions=output_attentions,
            use_cache=use_cache,
        )
        hidden_states = residual + hidden_states

        # Fully Connected
        residual = hidden_states
        hidden_states = self.post_attention_layernorm(hidden_states)
        hidden_states = self.mlp(hidden_states)
        hidden_states = residual + hidden_states

        outputs = (hidden_states,)

        if output_attentions:
            outputs += (self_attn_weights,)

        if use_cache:
            outputs += (present_key_value,)

        return outputs


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

    This model is also a PyTorch [torch.nn.Module](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.

    Parameters:
        config ([`XverseConfig`]):
            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
            [`~PreTrainedModel.from_pretrained`] method to load the model weights.
"""


@add_start_docstrings(
    "The bare Xverse Model outputting raw hidden-states without any specific head on top.",
    XVERSE_START_DOCSTRING,
)
class XversePreTrainedModel(PreTrainedModel):
    config_class = XverseConfig
    base_model_prefix = "model"
    supports_gradient_checkpointing = True
    _no_split_modules = ["XverseDecoderLayer"]
    _skip_keys_device_placement = "past_key_values"
    _keys_to_ignore_on_load_unexpected = [r"decoder\.version"]

    def _init_weights(self, module):
        std = self.config.initializer_range
        if isinstance(module, nn.Linear):
            module.weight.data.normal_(mean=0.0, std=std)
            if module.bias is not None:
                module.bias.data.zero_()
        elif isinstance(module, nn.Embedding):
            module.weight.data.normal_(mean=0.0, std=std)
            if module.padding_idx is not None:
                module.weight.data[module.padding_idx].zero_()

    def _set_gradient_checkpointing(self, module, value=False):
        if isinstance(module, XverseModel):
            module.gradient_checkpointing = value


XVERSE_INPUTS_DOCSTRING = r"""
    Args:
        input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
            Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you provide
            it.

            Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and
            [`PreTrainedTokenizer.__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)

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

            If `past_key_values` is used, optionally only the last `decoder_input_ids` have to be input (see
            `past_key_values`).

            If you want to change padding behavior, you should read [`modeling_opt._prepare_decoder_attention_mask`]
            and modify to your needs. See diagram 1 in [the paper](https://arxiv.org/abs/1910.13461) for more
            information on the default strategy.

            - 1 indicates the head is **not masked**,
            - 0 indicates the head is **masked**.
        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 (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
            Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2 tensors of shape
            `(batch_size, num_heads, sequence_length, embed_size_per_head)`) and 2 additional tensors of shape
            `(batch_size, num_heads, encoder_sequence_length, embed_size_per_head)`.

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

            If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids` (those that
            don't have their past key value states given to this model) of shape `(batch_size, 1)` instead of all
            `decoder_input_ids` of shape `(batch_size, sequence_length)`.
        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.
        use_cache (`bool`, *optional*):
            If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see
            `past_key_values`).
        output_attentions (`bool`, *optional*):
            Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
            tensors for more detail.
        output_hidden_states (`bool`, *optional*):
            Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
            more detail.
        return_dict (`bool`, *optional*):
            Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
"""


@add_start_docstrings(
    "The bare Xverse Model outputting raw hidden-states without any specific head on top.",
    XVERSE_START_DOCSTRING,
)
class XverseModel(XversePreTrainedModel):
    """
    Transformer decoder consisting of *config.num_hidden_layers* layers. Each layer is a [`XverseDecoderLayer`]

    Args:
        config: XverseConfig
    """

    def __init__(self, config: XverseConfig):
        super().__init__(config)
        self.padding_idx = config.pad_token_id
        self.vocab_size = config.vocab_size

        self.embed_tokens = nn.Embedding(
            config.vocab_size, config.hidden_size, self.padding_idx)
        self.layers = nn.ModuleList(
            [XverseDecoderLayer(config) for _ in range(config.num_hidden_layers)])
        self.norm = XverseRMSNorm(config.hidden_size, eps=config.rms_norm_eps)

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

    def get_input_embeddings(self):
        return self.embed_tokens

    def set_input_embeddings(self, value):
        self.embed_tokens = value

    # Copied from transformers.models.bart.modeling_bart.BartDecoder._prepare_decoder_attention_mask
    def _prepare_decoder_attention_mask(self, attention_mask, input_shape, inputs_embeds, past_key_values_length):
        # create causal mask
        # [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
        combined_attention_mask = None
        if input_shape[-1] > 1:
            combined_attention_mask = _make_causal_mask(
                input_shape,
                inputs_embeds.dtype,
                device=inputs_embeds.device,
                past_key_values_length=past_key_values_length,
            )

        if attention_mask is not None:
            # [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
            expanded_attn_mask = _expand_mask(attention_mask, inputs_embeds.dtype, tgt_len=input_shape[-1]).to(
                inputs_embeds.device
            )
            combined_attention_mask = (
                expanded_attn_mask if combined_attention_mask is None else expanded_attn_mask +
                combined_attention_mask
            )

        return combined_attention_mask

    @add_start_docstrings_to_model_forward(XVERSE_INPUTS_DOCSTRING)
    def forward(
        self,
        input_ids: torch.LongTensor = None,
        attention_mask: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        past_key_values: Optional[List[torch.FloatTensor]] = None,
        inputs_embeds: Optional[torch.FloatTensor] = None,
        use_cache: Optional[bool] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
    ) -> Union[Tuple, BaseModelOutputWithPast]:
        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

        # retrieve input_ids and inputs_embeds
        if input_ids is not None and inputs_embeds is not None:
            raise ValueError(
                "You cannot specify both decoder_input_ids and decoder_inputs_embeds at the same time")
        elif input_ids is not None:
            batch_size, seq_length = input_ids.shape
        elif inputs_embeds is not None:
            batch_size, seq_length, _ = inputs_embeds.shape
        else:
            raise ValueError(
                "You have to specify either decoder_input_ids or decoder_inputs_embeds")

        seq_length_with_past = seq_length
        past_key_values_length = 0

        if past_key_values is not None:
            past_key_values_length = past_key_values[0][0].shape[2]
            seq_length_with_past = seq_length_with_past + past_key_values_length

        if position_ids is None:
            device = input_ids.device if input_ids is not None else inputs_embeds.device
            position_ids = torch.arange(
                past_key_values_length, seq_length + past_key_values_length, dtype=torch.long, device=device
            )
            position_ids = position_ids.unsqueeze(0).view(-1, seq_length)
        else:
            position_ids = position_ids.view(-1, seq_length).long()

        if inputs_embeds is None:
            inputs_embeds = self.embed_tokens(input_ids)
        # embed positions
        if attention_mask is None:
            attention_mask = torch.ones(
                (batch_size, seq_length_with_past), dtype=torch.bool, device=inputs_embeds.device
            )
        attention_mask = self._prepare_decoder_attention_mask(
            attention_mask, (batch_size,
                             seq_length), inputs_embeds, past_key_values_length
        )

        hidden_states = inputs_embeds

        if self.gradient_checkpointing and self.training:
            if use_cache:
                logger.warning_once(
                    "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                )
                use_cache = False

        # decoder layers
        all_hidden_states = () if output_hidden_states else None
        all_self_attns = () if output_attentions else None
        next_decoder_cache = () if use_cache else None

        for idx, decoder_layer in enumerate(self.layers):
            if output_hidden_states:
                all_hidden_states += (hidden_states,)

            past_key_value = past_key_values[idx] if past_key_values is not None else None

            if self.gradient_checkpointing and self.training:

                def create_custom_forward(module):
                    def custom_forward(*inputs):
                        # None for past_key_value
                        return module(*inputs, output_attentions, None)

                    return custom_forward

                layer_outputs = torch.utils.checkpoint.checkpoint(
                    create_custom_forward(decoder_layer),
                    hidden_states,
                    attention_mask,
                    position_ids,
                    None,
                )
            else:
                layer_outputs = decoder_layer(
                    hidden_states,
                    attention_mask=attention_mask,
                    position_ids=position_ids,
                    past_key_value=past_key_value,
                    output_attentions=output_attentions,
                    use_cache=use_cache,
                )

            hidden_states = layer_outputs[0]

            if use_cache:
                next_decoder_cache += (
                    layer_outputs[2 if output_attentions else 1],)

            if output_attentions:
                all_self_attns += (layer_outputs[1],)

        hidden_states = self.norm(hidden_states)

        # add hidden states from the last decoder layer
        if output_hidden_states:
            all_hidden_states += (hidden_states,)

        next_cache = next_decoder_cache if use_cache else None
        if not return_dict:
            return tuple(v for v in [hidden_states, next_cache, all_hidden_states, all_self_attns] if v is not None)
        return BaseModelOutputWithPast(
            last_hidden_state=hidden_states,
            past_key_values=next_cache,
            hidden_states=all_hidden_states,
            attentions=all_self_attns,
        )


class XverseForCausalLM(XversePreTrainedModel):
    def __init__(self, config):
        super().__init__(config)
        self.model = XverseModel(config)

        self.lm_head = nn.Linear(
            config.hidden_size, config.vocab_size, bias=False)

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

    def get_input_embeddings(self):
        return self.model.embed_tokens

    def set_input_embeddings(self, value):
        self.model.embed_tokens = value

    def get_output_embeddings(self):
        return self.lm_head

    def set_output_embeddings(self, new_embeddings):
        self.lm_head = new_embeddings

    def set_decoder(self, decoder):
        self.model = decoder

    def get_decoder(self):
        return self.model

    @add_start_docstrings_to_model_forward(XVERSE_INPUTS_DOCSTRING)
    @replace_return_docstrings(output_type=CausalLMOutputWithPast, config_class=_CONFIG_FOR_DOC)
    def forward(
        self,
        input_ids: torch.LongTensor = None,
        attention_mask: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        past_key_values: Optional[List[torch.FloatTensor]] = None,
        inputs_embeds: Optional[torch.FloatTensor] = None,
        labels: Optional[torch.LongTensor] = None,
        use_cache: Optional[bool] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
    ) -> Union[Tuple, CausalLMOutputWithPast]:
        r"""
        Args:
            labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
                Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
                config.vocab_size]` or -100 (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:

        Example:

        ```python
        >>> from transformers import AutoTokenizer, AutoModelForCausalLM

        >>> model = AutoModelForCausalLM.from_pretrained(PATH_TO_CONVERTED_WEIGHTS, trust_remote_code=True)
        >>> tokenizer = AutoTokenizer.from_pretrained(PATH_TO_CONVERTED_TOKENIZER)

        >>> prompt = "Hey, are you conscious? Can you talk to me?"
        >>> inputs = tokenizer(prompt, return_tensors="pt")

        >>> # Generate
        >>> generate_ids = model.generate(inputs.input_ids, max_length=30)
        >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
        "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you."
        ```"""

        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.use_return_dict

        # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn)
        outputs = self.model(
            input_ids=input_ids,
            attention_mask=attention_mask,
            position_ids=position_ids,
            past_key_values=past_key_values,
            inputs_embeds=inputs_embeds,
            use_cache=use_cache,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )

        hidden_states = outputs[0]
        logits = self.lm_head(hidden_states)

        loss = None
        if labels is not None:
            # Shift so that tokens < n predict n
            shift_logits = logits[..., :-1, :].contiguous()
            shift_labels = labels[..., 1:].contiguous()
            # Flatten the tokens
            loss_fct = CrossEntropyLoss()
            shift_logits = shift_logits.view(-1, self.config.vocab_size)
            shift_labels = shift_labels.view(-1)
            # Enable model parallelism
            shift_labels = shift_labels.to(shift_logits.device)
            loss = loss_fct(shift_logits, shift_labels)

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

        return CausalLMOutputWithPast(
            loss=loss,
            logits=logits,
            past_key_values=outputs.past_key_values,
            hidden_states=outputs.hidden_states,
            attentions=outputs.attentions,
        )

    def _build_chat_input(self, tokenizer, messages: List[dict], max_new_tokens: int = 2048):
        max_new_tokens = max_new_tokens or self.generation_config.max_new_tokens
        max_input_tokens = self.config.max_position_embeddings - max_new_tokens
        max_input_tokens = max(
            self.config.max_position_embeddings // 2, max_input_tokens)
        max_input_tokens = min(
            self.config.max_tokenizer_truncation, max_input_tokens)

        total_input, round_input = [], []
        user_prompt_tokens = tokenizer.encode(
            "Human: ", return_token_type_ids=False)
        exec_prompt_tokens = tokenizer.encode(
            "Exec: ", return_token_type_ids=False)
        assist_prompt_tokens = tokenizer.encode(
            "Assistant: ", return_token_type_ids=False)
        assist_prompt_len = len(assist_prompt_tokens)

        for i, message in enumerate(messages[::-1]):
            if message['role'] == 'user' or message['role'] == 'exec':
                user_content = f"{message['content']}\n\n"
                content_tokens = user_prompt_tokens + tokenizer.encode(user_content, return_token_type_ids=False) if message['role'] == 'user' else \
                    exec_prompt_tokens + \
                    tokenizer.encode(user_content, return_token_type_ids=False)
                if i == 0:
                    content_tokens = content_tokens[:max_input_tokens -
                                                    assist_prompt_len]
                    content_tokens += assist_prompt_tokens
                round_input = content_tokens + round_input

                if i != 0:
                    if len(total_input) + len(round_input) > max_input_tokens:
                        break
                    else:
                        total_input = round_input + total_input
                else:
                    total_input = round_input + total_input
                    if len(total_input) >= max_input_tokens:
                        break
                round_input = []
            elif message['role'] == 'assistant':
                assist_content = f"{message['content']}"
                content_tokens = assist_prompt_tokens + \
                    tokenizer.encode(
                        assist_content, return_token_type_ids=False)
                round_input = content_tokens + \
                    [self.generation_config.eos_token_id] + round_input
            elif message['role'] == 'system':
                assert i == len(messages) - 1
                user_content = f"{message['content']}\n"
                content_tokens = tokenizer.encode(
                    user_content, return_token_type_ids=False)
                round_input = user_prompt_tokens + content_tokens + round_input
                if len(total_input) + len(round_input) > max_input_tokens:
                    break
                else:
                    total_input = round_input + total_input
            else:
                raise ValueError(
                    f"message role not supported yet: {message['role']}")
        total_input = torch.LongTensor([total_input]).to(self.device)
        return total_input

    @torch.no_grad()
    def chat(self, tokenizer, messages: List[dict], stream=False,
             generation_config: Optional[GenerationConfig] = None):
        generation_config = generation_config or self.generation_config
        input_ids = self._build_chat_input(
            tokenizer, messages, generation_config.max_new_tokens)
        if stream:
            from transformers import TextIteratorStreamer
            from threading import Thread
            streamer = TextIteratorStreamer(tokenizer, skip_prompt=True)
            self.__class__.generate = PreTrainedModel.generate

            def stream_generator():
                generation_kwargs = dict(
                    inputs=input_ids, generation_config=generation_config, streamer=streamer)
                thread = Thread(target=self.generate, kwargs=generation_kwargs)
                thread.start()
                for next_text in streamer:
                    yield next_text.replace(tokenizer.eos_token, "")

            return stream_generator()
        else:
            self.__class__.generate = PreTrainedModel.generate  # disable stream
            outputs = self.generate(
                input_ids, generation_config=generation_config)
            response = tokenizer.decode(
                outputs[0][len(input_ids[0]):], skip_special_tokens=True)
            return response

    def prepare_inputs_for_generation(
        self, input_ids, past_key_values=None, attention_mask=None, inputs_embeds=None, **kwargs
    ):
        if past_key_values:
            input_ids = input_ids[:, -1:]

        position_ids = kwargs.get("position_ids", None)
        if attention_mask is not None and position_ids is None:
            # create position_ids on the fly for batch generation
            position_ids = attention_mask.long().cumsum(-1) - 1
            position_ids.masked_fill_(attention_mask == 0, 1)
            if past_key_values:
                position_ids = position_ids[:, -1].unsqueeze(-1)

        # if `inputs_embeds` are passed, we only want to use them in the 1st generation step
        if inputs_embeds is not None and past_key_values is None:
            model_inputs = {"inputs_embeds": inputs_embeds}
        else:
            model_inputs = {"input_ids": input_ids}

        model_inputs.update(
            {
                "position_ids": position_ids,
                "past_key_values": past_key_values,
                "use_cache": kwargs.get("use_cache"),
                "attention_mask": attention_mask,
            }
        )
        return model_inputs

    @staticmethod
    def _reorder_cache(past_key_values, beam_idx):
        reordered_past = ()
        for layer_past in past_key_values:
            reordered_past += (tuple(past_state.index_select(0, beam_idx)
                               for past_state in layer_past),)
        return reordered_past

    def quantize(self, bit_length: int):
        from .quantization import QuantizationLinear

        for layer in self.model.layers:
            layer.self_attn.q_proj = QuantizationLinear(
                bit_length=bit_length,
                weight=layer.self_attn.q_proj.weight.to(
                    torch.cuda.current_device()),
                device=layer.self_attn.q_proj.weight.device,
            )
            layer.self_attn.k_proj = QuantizationLinear(
                bit_length=bit_length,
                weight=layer.self_attn.k_proj.weight.to(
                    torch.cuda.current_device()),
                device=layer.self_attn.k_proj.weight.device
            )
            layer.self_attn.v_proj = QuantizationLinear(
                bit_length=bit_length,
                weight=layer.self_attn.v_proj.weight.to(
                    torch.cuda.current_device()),
                device=layer.self_attn.v_proj.weight.device
            )
            layer.self_attn.o_proj = QuantizationLinear(
                bit_length=bit_length,
                weight=layer.self_attn.o_proj.weight.to(
                    torch.cuda.current_device()),
                device=layer.self_attn.o_proj.weight.device
            )
            layer.mlp.gate_proj = QuantizationLinear(
                bit_length=bit_length,
                weight=layer.mlp.gate_proj.weight.to(
                    torch.cuda.current_device()),
                device=layer.mlp.gate_proj.weight.device
            )
            layer.mlp.down_proj = QuantizationLinear(
                bit_length=bit_length,
                weight=layer.mlp.down_proj.weight.to(
                    torch.cuda.current_device()),
                device=layer.mlp.down_proj.weight.device
            )
            layer.mlp.up_proj = QuantizationLinear(
                bit_length=bit_length,
                weight=layer.mlp.up_proj.weight.to(
                    torch.cuda.current_device()),
                device=layer.mlp.up_proj.weight.device
            )

        return self