modeling_newgpt.py

from typing import Optional, Tuple, Union
import math

import numpy as np
from scipy.optimize import minimize

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

from transformers.activations import ACT2FN
from transformers.modeling_outputs import (
    BaseModelOutputWithPast,
    CausalLMOutputWithPast,
    QuestionAnsweringModelOutput,
    SequenceClassifierOutputWithPast,
)
from transformers.modeling_utils import PreTrainedModel
from transformers.utils import add_code_sample_docstrings, add_start_docstrings, add_start_docstrings_to_model_forward, logging
from transformers.utils.model_parallel_utils import assert_device_map, get_device_map
from .configuration_newgpt import NewGPTConfig



def fixed_pos_embedding(dim, seq_len, device):
    inv_freq = 1.0 / (10000 ** (torch.arange(0, dim, 2) / dim))
    sinusoid_inp = (
        torch.einsum("i , j -> i j", torch.arange(seq_len, dtype=torch.float), inv_freq).float().to(device)
    )
    return torch.sin(sinusoid_inp), torch.cos(sinusoid_inp)


def rotate_every_two(x):
    x1 = x[:, :, :, ::2]
    x2 = x[:, :, :, 1::2]
    x = torch.stack((-x2, x1), dim=-1)
    return x.flatten(-2)  # in einsum notation: rearrange(x, '... d j -> ... (d j)')

def get_scale(dim):
    x = np.arange(0, 4096, 1)
    inv_freq = 1.0 / (10000 ** (np.arange(0, dim, 2) / dim))
    angle = x[:, None] * inv_freq
    scale_init = -np.log(1 / ((np.arange(0, dim, 2) + 1 * dim) / (2 * dim)) - 1)
    def sigmoid(x):
        return 1 / (1 + np.exp(-x))

    def eval_fun(scale):
        scale = sigmoid(scale)
        posi_scale = (scale ** (x[:, None] / 256))
        upper_bound = (np.cos(angle) * posi_scale).mean(1) # \sum_i cos n\theta_i p^{n}
        delta = (upper_bound[:-1] - upper_bound[1:]) / upper_bound[:-1] # \sum_n (f(n) - f(n+1)) / f(n)
        return delta

    res = minimize(lambda scale: -eval_fun(scale).sum(), scale_init)
    return sigmoid(res.x)

def duplicate_interleave(m):
    """
    A simple version of `torch.repeat_interleave` for duplicating a matrix while interleaving the copy.
    """
    dim0 = m.shape[0]
    m = m.view(-1, 1)  # flatten the matrix
    m = m.repeat(1, 2)  # repeat all elements into the 2nd dimension
    m = m.view(dim0, -1)  # reshape into a matrix, interleaving the copy
    return m

def apply_rotary_pos_emb(x, sin, cos, offset=0, scale=1):
    sin, cos = map(lambda t: duplicate_interleave(t * scale)[offset : x.shape[2] + offset, :], (sin, cos))
    # einsum notation for lambda t: repeat(t[offset:x.shape[1]+offset,:], "n d -> () n () (d j)", j=2)
    return (x * cos) + (rotate_every_two(x) * sin)

def _make_causal_mask(
    input_ids_shape: torch.Size, device: torch.device, past_key_values_length: int) -> torch.BoolTensor:
    """
    Make causal mask used for self-attention.
    """
    batch_size, target_length = input_ids_shape
    mask = torch.empty((target_length, target_length + past_key_values_length), dtype=torch.bool, device=device)
    # ONNX doesn't support `torch.Tensor.triu` properly, thus we use this workaround
    seq_ids = torch.arange(target_length, device=device)
    mask[:, past_key_values_length:] = (seq_ids[:, None] < seq_ids[None, :])
    if past_key_values_length > 0:
        mask[:, :past_key_values_length] = False

    expanded_mask = mask[None, None, :, :].expand(batch_size, 1, target_length, target_length + past_key_values_length)
    return expanded_mask

def _expand_mask(mask: torch.Tensor, tgt_length: int) -> torch.BoolTensor:
    """
    Expands attention_mask from `[batch_size, src_length]` to `[batch_size, 1, tgt_length, src_length]`.
    """
    batch_size, src_length = mask.shape
    tgt_length = tgt_length if tgt_length is not None else src_length

    expanded_mask = ~(mask[:, None, None, :].to(torch.bool))
    return expanded_mask.expand(batch_size, 1, tgt_length, src_length)

def _split_heads(tensor, num_attention_heads, attn_head_size):
    """
    Splits hidden dim into attn_head_size and num_attention_heads
    """
    new_shape = tensor.size()[:-1] + (num_attention_heads, attn_head_size)
    tensor = tensor.view(new_shape)
    if len(tensor.shape) == 5:
        return tensor.permute(0, 1, 3, 2, 4)  # (batch, blocks, head, block_length, head_features)
    elif len(tensor.shape) == 4:
        return tensor.permute(0, 2, 1, 3)  # (batch, head, seq_length, head_features)
    elif len(tensor.shape) == 3:
        return tensor.permute(1, 0, 2)  # (head, seq_length, head_features)
    else:
        raise ValueError(f"Input tensor rank should be one of [4, 5], but is: {len(tensor.shape)}")

def _merge_heads(tensor, num_attention_heads, attn_head_size):
    """
    Merges attn_head_size dim and num_attn_heads dim into hidden dim
    """
    if len(tensor.shape) == 5:
        tensor = tensor.permute(0, 1, 3, 2, 4).contiguous()
    elif len(tensor.shape) == 4:
        tensor = tensor.permute(0, 2, 1, 3).contiguous()
    else:
        raise ValueError(f"Input tensor rank should be one of [4, 5], but is: {len(tensor.shape)}")
    new_shape = tensor.size()[:-2] + (num_attention_heads * attn_head_size,)
    return tensor.view(new_shape)

def build_alibi_tensor(attention_mask: torch.Tensor, num_heads: int, dtype: torch.dtype) -> torch.Tensor:
    """
    Link to paper: https://arxiv.org/abs/2108.12409 Alibi tensor is not causal as the original paper mentions, it
    relies on a translation invariance of softmax for quick implementation: with l being a tensor, and a fixed value
    `softmax(l+a) = softmax(l)`. Based on
    https://github.com/ofirpress/attention_with_linear_biases/blob/a35aaca144e0eb6b789dfcb46784c4b8e31b7983/fairseq/models/transformer.py#L742
    TODO @thomasw21 this doesn't work as nicely due to the masking strategy, and so masking varies slightly.

    Args:
    Returns tensor shaped (batch_size * num_heads, 1, max_seq_len)
        attention_mask (`torch.Tensor`):
            Token-wise attention mask, this should be of shape (batch_size, max_seq_len).
        num_heads (`int`, *required*):
            number of heads
        dtype (`torch.dtype`, *optional*, default=`torch.bfloat16`):
            dtype of the output tensor
    """
    batch_size, seq_length = attention_mask.shape

    closest_power_of_2 = 2 ** math.floor(math.log2(num_heads))
    base = torch.tensor(
        2 ** (-(2 ** -(math.log2(closest_power_of_2) - 3))), device=attention_mask.device, dtype=torch.float32
    )
    powers = torch.arange(1, 1 + closest_power_of_2, device=attention_mask.device, dtype=torch.int32)
    slopes = torch.pow(base, powers)

    if closest_power_of_2 != num_heads:
        extra_base = torch.tensor(
            2 ** (-(2 ** -(math.log2(2 * closest_power_of_2) - 3))), device=attention_mask.device, dtype=torch.float32
        )
        num_remaining_heads = min(closest_power_of_2, num_heads - closest_power_of_2)
        extra_powers = torch.arange(1, 1 + 2 * num_remaining_heads, 2, device=attention_mask.device, dtype=torch.int32)
        slopes = torch.cat([slopes, torch.pow(extra_base, extra_powers)], dim=0)

    # Note: alibi will added to the attention bias that will be applied to the query, key product of attention
    # => therefore alibi will have to be of shape (batch_size, num_heads, query_length, key_length)
    # => here we set (batch_size=1, num_heads=num_heads, query_length=1, key_length=max_length)
    # => the query_length dimension will then be broadcasted correctly
    # This is more or less identical to T5's relative position bias:
    # https://github.com/huggingface/transformers/blob/f681437203baa7671de3174b0fa583c349d9d5e1/src/transformers/models/t5/modeling_t5.py#L527
    arange_tensor = ((attention_mask.cumsum(dim=-1) - 1) * attention_mask)[:, None, :]
    alibi = slopes[..., None] * arange_tensor
    return alibi.reshape(batch_size, num_heads, 1, seq_length).to(dtype)

class NewGPTEMA(nn.Module):
    def __init__(self, config): 
        super().__init__()
        self.ndim = 8
        self.embed_dim = config.n_embd
        self.delta = nn.Parameter(torch.Tensor(self.embed_dim, self.ndim, 1))
        self.gamma = nn.Parameter(torch.Tensor(self.embed_dim, self.ndim, 1))
        self.scale = math.sqrt(1.0 / self.embed_dim)
        self.reset_parameters()

    def reset_parameters(self):
        nn.init.normal_(self.delta, mean=0.0, std=0.2)
        nn.init.normal_(self.gamma, mean=0.0, std=1.0)

    def build_kernel(self, length):
        p = torch.sigmoid(self.delta).float()
        kernel = torch.arange(length).to(p) * torch.log(p) # [embed_dim, ndim, length]
        kernel = torch.exp(kernel)
        kernel = torch.einsum('dnl,dn->dl', kernel, self.gamma.float().squeeze(-1) * self.scale)
        return kernel

    def forward(self, hidden_states: Optional[torch.FloatTensor]) -> torch.FloatTensor:
        # B x L x D -> B x nD x L
        length = hidden_states.shape[1]
        dtype = hidden_states.dtype
        hidden_states = hidden_states.permute(0, 2, 1)
        hidden_states = hidden_states.float()
         
        kernel = self.build_kernel(length)
        kernel = torch.fft.rfft(kernel, n=2 * length)
        hidden_states = torch.fft.rfft(hidden_states, n=2 * length)
        hidden_states = torch.fft.irfft(hidden_states * kernel, n=2 * length)[..., :length]
        # B x nD x L -> B x L x D
        hidden_states = hidden_states.permute(0, 2, 1)
        hidden_states = hidden_states.to(dtype)
        return hidden_states


class NewGPTJointAttention(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.resid_dropout = nn.Dropout(config.resid_pdrop)

        self.embed_dim = config.hidden_size
        self.num_attention_heads = config.num_attention_heads
        self.head_dim = self.embed_dim // self.num_attention_heads
        if self.head_dim * self.num_attention_heads != self.embed_dim:
            raise ValueError(
                f"embed_dim must be divisible by num_attention_heads (got `embed_dim`: {self.embed_dim} and"
                f" `num_attention_heads`: {self.num_attention_heads})."
            )
        self.scale_attn = torch.sqrt(torch.tensor(self.head_dim, dtype=torch.float32)).to(torch.get_default_dtype())
        self.q_proj = nn.Linear(self.embed_dim, self.embed_dim, bias=False)
        self.k_proj = nn.Linear(self.embed_dim, self.embed_dim, bias=False)
        self.v_proj = nn.Linear(self.embed_dim, self.embed_dim, bias=False)
        self.out_proj = nn.Linear(self.embed_dim, self.embed_dim, bias=False)

        # memory bias
        # self.memory_bias

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

        self.mode = config.mode
        if self.mode == "rot-momentum":
            self.ema = NewGPTEMA(config)

    def forward(
        self,
        hidden_states: Optional[torch.FloatTensor],
        attention_mask: Optional[torch.FloatTensor] = None,
        position_encoding: Optional[Tuple[torch.FloatTensor]] = None,
        layer_past: Optional[Tuple[torch.Tensor]] = None,
        head_mask: Optional[torch.FloatTensor] = None,
        use_cache: Optional[bool] = False,
        output_attentions: Optional[bool] = False,
        disable_add_index: Optional[bool] = False,
    ) -> Union[
        Tuple[torch.Tensor, Tuple[torch.Tensor]],
        Optional[Tuple[torch.Tensor, Tuple[torch.Tensor], Tuple[torch.Tensor, ...]]],
    ]:
        query = self.q_proj(hidden_states)
        key = self.k_proj(hidden_states)
        value = self.v_proj(hidden_states)

        if not external_memory:
            query = query.float() # (batch, seq_length, head_features)
            key = key.float()
            value = value.float()

        qkv_dict = {"q": query.detach(), "k": key.detach(), "v": value.detach()}

        seq_len = key.shape[1]
        offset = 0

        if layer_past is not None:
            offset = layer_past[0].shape[-2]
            seq_len += offset
            past_key = layer_past[0]
            past_value = layer_past[1]
            key = torch.cat((past_key, key), dim=-2)
            value = torch.cat((past_value, value), dim=-2)

        if use_cache is True:
            present = (key, value)
        else:
            present = None
        
        query = _split_heads(query, self.num_attention_heads, self.head_dim)
        key = _split_heads(key, self.num_attention_heads, self.head_dim)
        value_split = _split_heads(value, self.num_attention_heads, self.head_dim)
        if self.mode == "rotary" or self.mode == "rot-momentum":
            sin, cos = position_encoding
            key = apply_rotary_pos_emb(key, sin, cos)
            query = apply_rotary_pos_emb(query, sin, cos)
        elif self.mode == "rot-scale":
            sin, cos, scale = position_encoding
            key = apply_rotary_pos_emb(key, sin, cos, scale = 1 / scale)
            query = apply_rotary_pos_emb(query, sin, cos, scale = scale[(-query.shape[2]):])

        # retrieval to get keys and vals
        # query: bsz * nhead * seq_len * head_dim
        if external_memory:
            if external_memory.dstore_idx == 0:
                if not disable_add_index:
                    external_memory.add_index(qkv_dict)
            else:

                retrieval_output = external_memory.retrieve(qkv_dict['q'])
                long_context_retrieval = retrieval_output['tgt_index']
                
                retrieval_k, retrieval_v = long_context_retrieval['k'].to(query.device).type(query.dtype), long_context_retrieval['v'].to(query.device).type(query.dtype)

                attn_weights = torch.matmul(query, key.transpose(-1, -2)) / self.scale_attn
                attn_retrieval = torch.matmul(query.unsqueeze(-2), retrieval_k.transpose(-2, -1)).squeeze(-2) / self.scale_attn

                if self.mode == "alibi":
                    alibi = position_encoding[0]
                    attn_weights = attn_weights + alibi
                attn_weights = torch.masked_fill(attn_weights, attention_mask, torch.finfo(attn_weights.dtype).min)
                attn_text_probs = torch.softmax(attn_weights, dim=-1)
                attn_retrieval_probs = torch.softmax(attn_retrieval, dim=-1)

                attn_output = torch.matmul(attn_text_probs, value_split) * (1 - torch.sigmoid(self.memory_bias.reshape(1, self.num_attention_heads, 1, 1).repeat(key.shape[0], 1, 1, 1))) + torch.matmul(attn_retrieval_probs.unsqueeze(-2), retrieval_v).squeeze(-2) * torch.sigmoid(self.memory_bias.reshape(1, self.num_attention_heads, 1, 1))

                attn_output = _merge_heads(attn_output, self.num_attention_heads, self.head_dim)
                attn_output = attn_output.to(hidden_states.dtype)

                attn_output = self.out_proj(attn_output)
                attn_output = self.resid_dropout(attn_output)
                outputs = (attn_output, present)
                if output_attentions:
                    outputs += (attn_weights,)
                external_memory.add_index(qkv_dict)
                return outputs, qkv_dict

        # compute self-attention: V x Softmax(QK^T)  
        attn_weights = torch.matmul(query, key.transpose(-1, -2)) / self.scale_attn

        if self.mode == "alibi":
            alibi = position_encoding[0]
            attn_weights = attn_weights + alibi
            """
            Personal Notes:
            attn weights shape: bsz * nh * len * len
            alibi shape: bsz * nh * 1 * len
            """

        attn_weights = torch.masked_fill(attn_weights, attention_mask, torch.finfo(attn_weights.dtype).min)
        attn_weights = nn.functional.softmax(attn_weights, dim=-1)

        attn_output = torch.matmul(attn_weights, value_split)
        attn_output = _merge_heads(attn_output, self.num_attention_heads, self.head_dim)

        if self.mode == "rot-momentum":
            ema_output = self.ema(value)
            attn_output = attn_output + ema_output

        attn_output = attn_output.to(hidden_states.dtype)
        attn_output = self.out_proj(attn_output)
        attn_output = self.resid_dropout(attn_output)
        outputs = (attn_output, present)
        if output_attentions:
            outputs += (attn_weights,)

        return outputs, qkv_dict  # a, present, (attentions)



class NewGPTAttention(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.resid_dropout = nn.Dropout(config.resid_pdrop)

        self.embed_dim = config.hidden_size
        self.num_attention_heads = config.num_attention_heads
        self.head_dim = self.embed_dim // self.num_attention_heads
        if self.head_dim * self.num_attention_heads != self.embed_dim:
            raise ValueError(
                f"embed_dim must be divisible by num_attention_heads (got `embed_dim`: {self.embed_dim} and"
                f" `num_attention_heads`: {self.num_attention_heads})."
            )
        self.scale_attn = torch.sqrt(torch.tensor(self.head_dim, dtype=torch.float32)).to(torch.get_default_dtype())
        self.q_proj = nn.Linear(self.embed_dim, self.embed_dim, bias=False)
        self.k_proj = nn.Linear(self.embed_dim, self.embed_dim, bias=False)
        self.v_proj = nn.Linear(self.embed_dim, self.embed_dim, bias=False)
        self.out_proj = nn.Linear(self.embed_dim, self.embed_dim, bias=False)

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

        self.mode = config.mode
        if self.mode == "rot-momentum":
            self.ema = NewGPTEMA(config)

    def forward(
        self,
        hidden_states: Optional[torch.FloatTensor],
        attention_mask: Optional[torch.FloatTensor] = None,
        position_encoding: Optional[Tuple[torch.FloatTensor]] = None,
        layer_past: Optional[Tuple[torch.Tensor]] = None,
        head_mask: Optional[torch.FloatTensor] = None,
        use_cache: Optional[bool] = False,
        output_attentions: Optional[bool] = False,
        disable_add_index: Optional[bool] = False,
    ) -> Union[
        Tuple[torch.Tensor, Tuple[torch.Tensor]],
        Optional[Tuple[torch.Tensor, Tuple[torch.Tensor], Tuple[torch.Tensor, ...]]],
    ]:
        query = self.q_proj(hidden_states)
        key = self.k_proj(hidden_states)
        value = self.v_proj(hidden_states)

        query = query.float() # (batch, seq_length, head_features)
        key = key.float()
        value = value.float()

        qkv_dict = {"q": query, "k": key, "v": value}

        seq_len = key.shape[1]
        offset = 0

        if layer_past is not None:
            offset = layer_past[0].shape[-2]
            seq_len += offset
            past_key = layer_past[0]
            past_value = layer_past[1]
            key = torch.cat((past_key, key), dim=-2)
            value = torch.cat((past_value, value), dim=-2)

        if use_cache is True:
            present = (key, value)
        else:
            present = None
          
        query = _split_heads(query, self.num_attention_heads, self.head_dim)
        key = _split_heads(key, self.num_attention_heads, self.head_dim)
        value_split = _split_heads(value, self.num_attention_heads, self.head_dim)
        if self.mode == "rotary" or self.mode == "rot-momentum":
            sin, cos = position_encoding
            key = apply_rotary_pos_emb(key, sin, cos)
            query = apply_rotary_pos_emb(query, sin, cos)
        elif self.mode == "rot-scale":
            sin, cos, scale = position_encoding
            key = apply_rotary_pos_emb(key, sin, cos, scale = 1 / scale)
            query = apply_rotary_pos_emb(query, sin, cos, scale = scale[(-query.shape[2]):])

        # compute self-attention: V x Softmax(QK^T)  
        attn_weights = torch.matmul(query, key.transpose(-1, -2)) / self.scale_attn
        if self.mode == "alibi":
            alibi = position_encoding[0]
            attn_weights = attn_weights + alibi
            """
            Personal Notes:
            attn weights shape: bsz * nh * len * len
            alibi shape: bsz * nh * 1 * len
            """

        attn_weights = torch.masked_fill(attn_weights, attention_mask, torch.finfo(attn_weights.dtype).min)
        attn_weights = nn.functional.softmax(attn_weights, dim=-1)

        attn_output = torch.matmul(attn_weights, value_split)
        attn_output = _merge_heads(attn_output, self.num_attention_heads, self.head_dim)

        if self.mode == "rot-momentum":
            ema_output = self.ema(value)
            attn_output = attn_output + ema_output

        attn_output = attn_output.to(hidden_states.dtype)
        attn_output = self.out_proj(attn_output)
        attn_output = self.resid_dropout(attn_output)
        outputs = (attn_output, present)
        if output_attentions:
            outputs += (attn_weights,)

        return outputs, qkv_dict  # a, present, (attentions)


class NewGPTMLP(nn.Module):
    def __init__(self, intermediate_size, config):  # in MLP: intermediate_size= 4 * embed_dim
        super().__init__()
        embed_dim = config.n_embd

        self.fc_in = nn.Linear(embed_dim, intermediate_size)
        self.fc_out = nn.Linear(intermediate_size, embed_dim)

        self.act = ACT2FN[config.activation_function]
        self.dropout = nn.Dropout(config.resid_pdrop)

    def forward(self, hidden_states: Optional[torch.FloatTensor]) -> torch.FloatTensor:
        hidden_states = self.fc_in(hidden_states)
        hidden_states = self.act(hidden_states)
        hidden_states = self.fc_out(hidden_states)
        hidden_states = self.dropout(hidden_states)
        return hidden_states


class NewGPTBlock(nn.Module):
    def __init__(self, config):
        super().__init__()
        inner_dim = config.n_inner if config.n_inner is not None else 4 * config.n_embd
        self.ln_1 = nn.LayerNorm(config.n_embd, eps=config.layer_norm_epsilon)
        self.attn = NewGPTAttention(config)
        self.mlp = NewGPTMLP(inner_dim, config)

    def forward(
        self,
        hidden_states: Optional[torch.FloatTensor],
        layer_past: Optional[Tuple[torch.Tensor]] = None,
        attention_mask: Optional[torch.FloatTensor] = None,
        position_encoding: Optional[Tuple[torch.FloatTensor]] = None,
        head_mask: Optional[torch.FloatTensor] = None,
        use_cache: Optional[bool] = False,
        output_attentions: Optional[bool] = False,
        disable_add_index: Optional[bool] = False,
    ) -> Union[Tuple[torch.Tensor], Optional[Tuple[torch.Tensor, Tuple[torch.FloatTensor, ...]]]]:
        residual = hidden_states
        hidden_states = self.ln_1(hidden_states)
        attn_outputs, qkv_dict = self.attn(
            hidden_states,
            layer_past=layer_past,
            attention_mask=attention_mask,
            position_encoding=position_encoding,
            head_mask=head_mask,
            use_cache=use_cache,
            output_attentions=output_attentions,
            disable_add_index=disable_add_index,
        )
        attn_output = attn_outputs[0]  # output_attn: a, present, (attentions)
        outputs = attn_outputs[1:]

        feed_forward_hidden_states = self.mlp(hidden_states)
        hidden_states = attn_output + feed_forward_hidden_states + residual
        
        if use_cache:
            outputs = (hidden_states,) + outputs
        else:
            outputs = (hidden_states,) + outputs[1:]

        return outputs, qkv_dict  # hidden_states, present, (attentions)


class NewGPTPreTrainedModel(PreTrainedModel):
    """
    An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained
    models.
    """

    config_class = NewGPTConfig
    base_model_prefix = "transformer"
    is_parallelizable = True
    supports_gradient_checkpointing = True
    _no_split_modules = ["NewGPTBlock"]

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

    def _init_weights(self, module):
        """Initialize the weights."""
        if isinstance(module, (nn.Linear,)):
            # Slightly different from Mesh Transformer JAX 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)

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


class NewGPTModel(NewGPTPreTrainedModel):
    def __init__(self, config):
        super().__init__(config)

        self.mode = config.mode
        self.embed_dim = config.n_embd
        self.num_attention_heads = config.num_attention_heads
        self.vocab_size = config.vocab_size
        self.wte = nn.Embedding(config.vocab_size, self.embed_dim)
        self.drop = nn.Dropout(config.embd_pdrop)
        self.h = nn.ModuleList([NewGPTBlock(config) for _ in range(config.n_layer)])
        if config.use_external_memory:
            self.h[config.retrieval_layer_index].attn = NewGPTJointAttention(config)
        self.ln_f = nn.LayerNorm(self.embed_dim, eps=config.layer_norm_epsilon)
        
        # print(config)
        self.retrieval_layer_index = getattr(config, "retrieval_layer_index", 17)
        print("NewGPT retrieval Layer Index", self.retrieval_layer_index)
        # Model parallel
        self.model_parallel = False
        self.device_map = None
        self.gradient_checkpointing = False

        self.head_dim = self.embed_dim // self.num_attention_heads
        self.scale_div = 128
        if self.mode == "rot-scale":
            self.scale_base = get_scale(self.head_dim)
            
        # Initialize weights and apply final processing
        self.post_init()

    def parallelize(self, device_map=None):
        # Check validity of device_map
        self.device_map = (
            get_device_map(len(self.h), range(torch.cuda.device_count())) if device_map is None else device_map
        )
        assert_device_map(self.device_map, len(self.h))
        self.model_parallel = True
        self.first_device = "cpu" if "cpu" in self.device_map.keys() else "cuda:" + str(min(self.device_map.keys()))
        self.last_device = "cuda:" + str(max(self.device_map.keys()))
        self.wte = self.wte.to(self.first_device)
        # Load onto devices
        for k, v in self.device_map.items():
            for block in v:
                cuda_device = "cuda:" + str(k)
                self.h[block] = self.h[block].to(cuda_device)
        # ln_f to last
        self.ln_f = self.ln_f.to(self.last_device)

    def deparallelize(self):
        self.model_parallel = False
        self.device_map = None
        self.first_device = "cpu"
        self.last_device = "cpu"
        self.wte = self.wte.to("cpu")
        for index in range(len(self.h)):
            self.h[index] = self.h[index].to("cpu")
        self.ln_f = self.ln_f.to("cpu")
        torch.cuda.empty_cache()

    def get_input_embeddings(self):
        return self.wte

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

    def _prepare_attn_mask(
        self, attention_mask: torch.Tensor, input_shape: Tuple[int, int], past_key_values_length: int
    ) -> torch.BoolTensor:
        # create causal mask
        # [batch_size, seq_length] -> [batch_size, 1, tgt_length, src_length]
        combined_attention_mask = None
        device = attention_mask.device
        _, src_length = input_shape

        if src_length > 1:
            combined_attention_mask = _make_causal_mask(
                input_shape, device=device, past_key_values_length=past_key_values_length)

        # [batch_size, seq_length] -> [batch_size, 1, tgt_length, src_length]
        expanded_attn_mask = _expand_mask(attention_mask, tgt_length=src_length)
        combined_attention_mask = (
            expanded_attn_mask if combined_attention_mask is None else expanded_attn_mask | combined_attention_mask
        )

        return combined_attention_mask
    

    def forward(
        self,
        input_ids: Optional[torch.LongTensor] = None,
        past_key_values: Optional[Tuple[Tuple[torch.Tensor]]] = None,
        attention_mask: Optional[torch.FloatTensor] = None,
        token_type_ids: Optional[torch.LongTensor] = None,
        head_mask: Optional[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,
        disable_add_index: Optional[bool] = False,
    ) -> 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

        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:
            input_shape = input_ids.size()
            input_ids = input_ids.view(-1, input_shape[-1])
            inputs_embeds = self.wte(input_ids)
            batch_size = input_ids.shape[0]
        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")

        device = input_ids.device if input_ids is not None else inputs_embeds.device
        batch_size = inputs_embeds.shape[0]
        if token_type_ids is not None:
            token_type_ids = token_type_ids.view(-1, input_shape[-1])

        if past_key_values is None:
            past_length = 0
            past_key_values = tuple([None] * len(self.h))
        else:
            past_length = past_key_values[0][0].size(-2)

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

        hidden_states = inputs_embeds

        if token_type_ids is not None:
            token_type_embeds = self.wte(token_type_ids)
            hidden_states = hidden_states + token_type_embeds

        total_length = input_shape[-1] + past_length
        if self.mode == "absolute":
            sin, cos = fixed_pos_embedding(self.embed_dim, total_length, hidden_states.device)
            wpe = torch.stack((sin, cos), dim=-1).flatten(-2).to(hidden_states.dtype)
            hidden_states = hidden_states + wpe[past_length:]

        hidden_states = self.drop(hidden_states)

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

        if attention_mask is None:
            attention_mask = torch.ones((batch_size, input_shape[-1] + past_length), device=hidden_states.device)
        else:
            attention_mask = attention_mask.view(-1, input_shape[-1]).to(hidden_states.device)
        
        if self.mode == "alibi":
            alibi = build_alibi_tensor(attention_mask, self.num_attention_heads, dtype=hidden_states.dtype)
            position_encoding = (alibi,)
        elif self.mode == "rotary" or self.mode == "rot-scale" or self.mode == "rot-momentum":
            position_encoding = fixed_pos_embedding(self.head_dim, total_length, hidden_states.device)
            if self.mode == "rot-scale":
                scale = torch.from_numpy(self.scale_base).float() ** torch.arange(0, total_length, 1).div(self.scale_div)[:, None]
                position_encoding += (scale.to(hidden_states.device),)
        else:
            position_encoding = None
        
        attention_mask = self._prepare_attn_mask(
            attention_mask,
            input_shape=(batch_size, input_shape[-1]),
            past_key_values_length=past_length,
        )

        presents = () if use_cache else None
        all_self_attentions = () if output_attentions else None
        all_hidden_states = () if output_hidden_states else None
        return_kv_dict = None
        for i, (block, layer_past) in enumerate(zip(self.h, past_key_values)):

            # Model parallel
            if self.model_parallel:
                torch.cuda.set_device(hidden_states.device)
                # Ensure layer_past is on same device as hidden_states (might not be correct)
                if layer_past is not None:
                    layer_past = tuple(past_state.to(hidden_states.device) for past_state in layer_past)
                # Ensure that attention_mask is always on the same device as hidden_states
                attention_mask = attention_mask.to(hidden_states.device)
                if isinstance(head_mask, torch.Tensor):
                    head_mask = head_mask.to(hidden_states.device)
            if output_hidden_states:
                all_hidden_states = all_hidden_states + (hidden_states,)

            if self.gradient_checkpointing and self.training:

                if use_cache:
                    use_cache = False

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

                    return custom_forward

                outputs = torch.utils.checkpoint.checkpoint(
                    create_custom_forward(block),
                    hidden_states,
                    None,
                    attention_mask,
                    head_mask[i],
                )
            else:
                outputs, qkv_dict = block(
                    hidden_states,
                    layer_past=layer_past,
                    attention_mask=attention_mask,
                    position_encoding=position_encoding,
                    head_mask=head_mask[i],
                    use_cache=use_cache,
                    output_attentions=output_attentions,
                    disable_add_index=disable_add_index,
                )
                if i == self.retrieval_layer_index:
                    # print(self.retrieval_layer_index)
                    return_qkv_dict = qkv_dict

            hidden_states = outputs[0]
            if use_cache is True:
                presents = presents + (outputs[1],)

            if output_attentions:
                all_self_attentions = all_self_attentions + (outputs[2 if use_cache else 1],)

            # Model Parallel: If it's the last layer for that device, put things on the next device
            if self.model_parallel:
                for k, v in self.device_map.items():
                    if i == v[-1] and "cuda:" + str(k) != self.last_device:
                        hidden_states = hidden_states.to("cuda:" + str(k + 1))

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

        return BaseModelOutputWithPast(
            last_hidden_state=hidden_states,
            past_key_values=presents,
            hidden_states=all_hidden_states,
            attentions=all_self_attentions,
        ), return_qkv_dict, position_encoding


class NewGPTForCausalLM(NewGPTPreTrainedModel):
    _keys_to_ignore_on_load_missing = [r"h\.\d+\.attn\.masked_bias", r"h\.\d+\.attn\.bias"]

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

        # Model parallel
        self.model_parallel = False
        self.device_map = None

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

    def parallelize(self, device_map=None):
        self.device_map = (
            get_device_map(len(self.transformer.h), range(torch.cuda.device_count()))
            if device_map is None
            else device_map
        )
        assert_device_map(self.device_map, len(self.transformer.h))
        self.transformer.parallelize(self.device_map)
        self.lm_head = self.lm_head.to(self.transformer.first_device)
        self.model_parallel = True

    def deparallelize(self):
        self.transformer.deparallelize()
        self.transformer = self.transformer.to("cpu")
        self.lm_head = self.lm_head.to("cpu")
        self.model_parallel = False
        torch.cuda.empty_cache()

    def get_output_embeddings(self):
        return self.lm_head

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

    def prepare_inputs_for_generation(self, input_ids, past=None, **kwargs):
        token_type_ids = kwargs.get("token_type_ids", None)
        # only last token for inputs_ids if past is defined in kwargs
        if past:
            input_ids = input_ids[:, -1].unsqueeze(-1)
            if token_type_ids is not None:
                token_type_ids = token_type_ids[:, -1].unsqueeze(-1)

        attention_mask = kwargs.get("attention_mask", None)
        return {
            "input_ids": input_ids,
            "past_key_values": past,
            "use_cache": kwargs.get("use_cache"),
            "attention_mask": attention_mask,
            "token_type_ids": token_type_ids,
        }

    def forward(
        self,
        input_ids: Optional[torch.LongTensor] = None,
        past_key_values: Optional[Tuple[Tuple[torch.Tensor]]] = None,
        attention_mask: Optional[torch.FloatTensor] = None,
        token_type_ids: Optional[torch.LongTensor] = None,
        head_mask: Optional[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"""
        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.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,
            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]

        # Set device for model parallelism
        if self.model_parallel:
            torch.cuda.set_device(self.transformer.first_device)
            hidden_states = hidden_states.to(self.lm_head.weight.device)

        # make sure sampling in fp16 works correctly and
        # compute loss in fp32 to match with mesh-tf version
        # https://github.com/EleutherAI/gpt-neo/blob/89ce74164da2fb16179106f54e2269b5da8db333/models/gpt2/gpt2.py#L179
        lm_logits = self.lm_head(hidden_states).to(torch.float32)

        loss = None
        if labels is not None:
            # Shift so that tokens < n predict n
            shift_logits = lm_logits[..., :-1, :].contiguous()
            shift_labels = labels[..., 1:].contiguous()
            # Flatten the tokens
            loss_fct = CrossEntropyLoss()
            loss = loss_fct(shift_logits.view(-1, shift_logits.size(-1)), shift_labels.view(-1))

            loss = loss.to(hidden_states.dtype)

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

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

    @staticmethod
    def _reorder_cache(past: Tuple[Tuple[torch.Tensor]], beam_idx: torch.Tensor) -> Tuple[Tuple[torch.Tensor]]:
        """
        This function is used to re-order the `past_key_values` cache if [`~PretrainedModel.beam_search`] or
        [`~PretrainedModel.beam_sample`] is called. This is required to match `past_key_values` with the correct
        beam_idx at every generation step.
        """
        return tuple(
            tuple(past_state.index_select(0, beam_idx.to(past_state.device)) for past_state in layer_past)
            for layer_past in past
        )


class NewGPTForSequenceClassification(NewGPTPreTrainedModel):
    _keys_to_ignore_on_load_missing = [r"h\.\d+\.attn\.masked_bias", r"h\.\d+\.attn\.bias", r"lm_head.weight"]

    def __init__(self, config):
        super().__init__(config)
        self.num_labels = config.num_labels
        self.transformer = NewGPTModel(config)
        self.score = nn.Linear(config.n_embd, self.num_labels, bias=False)

        # Model parallel
        self.model_parallel = False
        self.device_map = None

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

    def forward(
        self,
        input_ids: Optional[torch.LongTensor] = None,
        past_key_values: Optional[Tuple[Tuple[torch.Tensor]]] = None,
        attention_mask: Optional[torch.FloatTensor] = None,
        token_type_ids: Optional[torch.LongTensor] = None,
        head_mask: Optional[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, SequenceClassifierOutputWithPast]:
        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.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,
            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]
        logits = self.score(hidden_states)

        if input_ids is not None:
            batch_size = input_ids.shape[0]
        else:
            batch_size = inputs_embeds.shape[0]

        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:
            sequence_lengths = -1
        else:
            if input_ids is not None:
                sequence_lengths = torch.ne(input_ids, self.config.pad_token_id).sum(-1) - 1
            else:
                sequence_lengths = -1

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

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


class NewGPTForQuestionAnswering(NewGPTPreTrainedModel):
    _keys_to_ignore_on_load_missing = [r"h\.\d+\.attn\.masked_bias", r"h\.\d+\.attn\.bias", r"lm_head.weight"]

    def __init__(self, config):
        super().__init__(config)
        self.num_labels = config.num_labels
        self.transformer = NewGPTModel(config)
        self.qa_outputs = nn.Linear(config.hidden_size, config.num_labels)

        # Model parallel
        self.model_parallel = False
        self.device_map = None

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

    def forward(
        self,
        input_ids: Optional[torch.LongTensor] = None,
        attention_mask: Optional[torch.FloatTensor] = None,
        token_type_ids: Optional[torch.LongTensor] = None,
        head_mask: Optional[torch.FloatTensor] = None,
        inputs_embeds: Optional[torch.FloatTensor] = None,
        start_positions: Optional[torch.LongTensor] = None,
        end_positions: Optional[torch.LongTensor] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
    ) -> Union[Tuple, QuestionAnsweringModelOutput]:
        r"""
        start_positions (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
            Labels for position (index) of the start of the labelled span for computing the token classification loss.
            Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the sequence
            are not taken into account for computing the loss.
        end_positions (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
            Labels for position (index) of the end of the labelled span for computing the token classification loss.
            Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the sequence
            are not taken into account for computing the loss.
        """
        return_dict = return_dict if return_dict is not None else self.config.use_return_dict

        outputs = self.transformer(
            input_ids,
            attention_mask=attention_mask,
            token_type_ids=token_type_ids,
            head_mask=head_mask,
            inputs_embeds=inputs_embeds,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )

        sequence_output = outputs[0]

        logits = self.qa_outputs(sequence_output)
        start_logits, end_logits = logits.split(1, dim=-1)
        start_logits = start_logits.squeeze(-1).contiguous()
        end_logits = end_logits.squeeze(-1).contiguous()

        total_loss = None
        if start_positions is not None and end_positions is not None:
            # If we are on multi-GPU, split add a dimension
            if len(start_positions.size()) > 1:
                start_positions = start_positions.squeeze(-1)
            if len(end_positions.size()) > 1:
                end_positions = end_positions.squeeze(-1)
            # sometimes the start/end positions are outside our model inputs, we ignore these terms
            ignored_index = start_logits.size(1)
            start_positions = start_positions.clamp(0, ignored_index)
            end_positions = end_positions.clamp(0, ignored_index)

            loss_fct = CrossEntropyLoss(ignore_index=ignored_index)
            start_loss = loss_fct(start_logits, start_positions)
            end_loss = loss_fct(end_logits, end_positions)
            total_loss = (start_loss + end_loss) / 2

        if not return_dict:
            output = (start_logits, end_logits) + outputs[2:]
            return ((total_loss,) + output) if total_loss is not None else output

        return QuestionAnsweringModelOutput(
            loss=total_loss,
            start_logits=start_logits,
            end_logits=end_logits,
            hidden_states=outputs.hidden_states,
            attentions=outputs.attentions,
        )