modeling_backpack_gpt2.py

import math
from dataclasses import dataclass
from typing import Optional, Tuple

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

from transformers.activations import ACT2FN
from transformers.pytorch_utils import Conv1D
from transformers.utils import (
    ModelOutput,
    logging,
)
from transformers.models.gpt2.modeling_gpt2 import GPT2Model, GPT2PreTrainedModel, GenerationMixin
from transformers.cache_utils import Cache
from .configuration_backpack_gpt2 import BackpackGPT2Config

logger = logging.get_logger(__name__)


### Backpack-Specific
class BackpackGPT2PreTrainedModel(GPT2PreTrainedModel):
    """
    An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained
    models.
    """
    _keys_to_ignore_on_load_missing = [r"attn.masked_bias", r"attn.bias"]

    config_class = BackpackGPT2Config
    base_model_prefix = "backpack"
    is_parallelizable = True
    supports_gradient_checkpointing = False
    _no_split_modules = ["GPT2Block", "BackpackNoMixBlock"]

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

class BackpackMLP(nn.Module):

  def __init__(self, embed_dim, intermediate_dim, out_dim, config):
        super().__init__()
        self.c_fc = Conv1D(intermediate_dim, embed_dim)
        self.c_proj = Conv1D(out_dim, intermediate_dim)
        self.act = ACT2FN[config.activation_function]
        self.dropout = nn.Dropout(config.resid_pdrop)

  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 BackpackNoMixBlock(nn.Module):

  def __init__(self, config):
    super().__init__()
    self.ln_1 = nn.LayerNorm(config.n_embd, eps=config.layer_norm_epsilon)
    self.ln_2 = nn.LayerNorm(config.n_embd, eps=config.layer_norm_epsilon)
    self.mlp = BackpackMLP(config.n_embd, config.n_embd*4, config.n_embd, config)
    self.resid_dropout1 = nn.Dropout(config.resid_pdrop)
    self.resid_dropout2 = nn.Dropout(config.resid_pdrop)

  def forward(self, hidden_states, residual):
    residual = self.resid_dropout1(hidden_states) + residual
    hidden_states = self.ln_1(residual)
    mlp_out = self.mlp(hidden_states)
    residual = self.resid_dropout2(mlp_out) + residual
    hidden_states = self.ln_2(residual)
    return hidden_states


class BackpackSenseNetwork(nn.Module):
    def __init__(self, config, num_senses, device=None, dtype=None):
        super().__init__()
        self.num_senses = num_senses
        #self.embeddings = embeddings
        self.n_embd = config.n_embd

        self.dropout = nn.Dropout(config.embd_pdrop)
        self.block = BackpackNoMixBlock(config)
        self.ln = nn.LayerNorm(self.n_embd, eps=config.layer_norm_epsilon)
        self.final_mlp = BackpackMLP(
            embed_dim=config.n_embd,
            intermediate_dim=config.sense_intermediate_scale*config.n_embd,
            out_dim=config.n_embd*config.num_senses,
            config=config,
            )

    def forward(self, input_embeds):
      residual = self.dropout(input_embeds)
      hidden_states = self.ln(residual)
      hidden_states = self.block(hidden_states, residual)
      senses = self.final_mlp(hidden_states)
      bs, s, nvd = senses.shape
      return senses.reshape(bs, s, self.num_senses, self.n_embd).transpose(1,2) # (bs, nv, s, d)

class BackpackWeightNetwork(nn.Module):

  def __init__(self, num_senses, embed_dim):
    super().__init__()
    self.n_embd = embed_dim
    self.num_senses = num_senses
    self.embed_per_sense = embed_dim // num_senses
    self.c_attn = nn.Linear(embed_dim, 2 * num_senses * self.embed_per_sense)
    self.softmax_scale = None

  def forward(self, encoded):
    """
    b, s, d = encoded.shape
    encoded = self.c_attn(encoded) # (b, s, 2*d)
    encoded = encoded.reshape(b, s, 2, self.num_senses, self.embed_per_sense) #(b, s, 2, nv, d//nv)
    batch_size, seqlen = encoded.shape[0], encoded.shape[1]

    # compute scores & mask
    q, k = encoded.unbind(dim=2)
    softmax_scale = self.softmax_scale or 1.0 / math.sqrt(q.shape[-1])
    scores = torch.einsum('bthd,bshd->bhts', q, k * softmax_scale)
    causal_mask = torch.triu(torch.full((seqlen, seqlen), -10000.0, device=scores.device), 1)
    scores = scores + causal_mask.to(dtype=scores.dtype)

    return torch.softmax(scores, dim=-1, dtype=q.dtype)
    """
    b, s, d = encoded.shape
    x = self.c_attn(encoded)  # (b, s, 2*d)
    x = x.reshape(b, s, 2, self.num_senses, self.embed_per_sense)  # (b, s, 2, nv, d//nv)

    # q,k: (b, s, nv, d//nv)
    q, k = x.unbind(dim=2)

    # scale (compute as float32 to reduce rounding error, then cast)
    scale = (self.softmax_scale
             if self.softmax_scale is not None
             else 1.0 / math.sqrt(q.shape[-1]))
    # einsum gives (b, nv, s, s)
    scores = torch.einsum('bthd,bshd->bhts', q, k) * scale  # keep native dtype here

    # boolean causal mask: True = mask-out
    # shape (s, s) → broadcast to (1, 1, s, s) → (b, nv, s, s)
    causal_mask = torch.ones(s, s, device=scores.device, dtype=torch.bool).triu_(1)
    scores = scores.float()  # do the numerically sensitive bits in fp32
    scores = scores.masked_fill(causal_mask, float('-inf'))

    attn = torch.softmax(scores, dim=-1)          # fp32 softmax
    attn = attn.to(q.dtype)                       # cast back (fp16/bf16) for downstream

    return attn

@dataclass
class BackpackGPT2BaseModelOutput(ModelOutput):
    hidden_states: torch.FloatTensor = None
    contextualization: torch.FloatTensor = None
    senses: torch.FloatTensor = None

class BackpackGPT2Model(BackpackGPT2PreTrainedModel):
    _keys_to_ignore_on_load_missing = [r".*attn.masked_bias", r".*attn.bias"]

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

        self.embed_dim = config.n_embd

        self.num_senses = config.num_senses
        self.gpt2_model = GPT2Model(config)
        self.sense_network = BackpackSenseNetwork(config, self.num_senses, self.gpt2_model.wte)
        self.word_embeddings = self.gpt2_model.wte
        self.position_embeddings = self.gpt2_model.wpe
        self.sense_weight_net = BackpackWeightNetwork(self.num_senses, self.embed_dim)
        # Model parallel
        self.model_parallel = False
        self.device_map = None
        self.gradient_checkpointing = False

    def get_num_senses(self):
        return self.num_senses

    def get_word_embeddings(self):
        return self.word_embeddings

    def get_sense_network(self):
        return self.sense_network

    def get_input_embeddings(self):
        return self.word_embeddings

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

    def forward(
        self, 
        input_ids, 
        position_ids,
        cache_position: Optional[torch.LongTensor] = None,
        past_key_values: Optional[Cache] = None,
        inputs_embeds: Optional[torch.FloatTensor] = None,
        attention_mask: Optional[torch.FloatTensor] = None,
        use_cache: Optional[bool] = None,
        return_dict: Optional[bool] = None,
        **kwargs):
        # Compute senses
        sense_input_embeds = self.word_embeddings(input_ids)
        senses = self.sense_network(sense_input_embeds) # (bs, nv, s, d)

        # Compute contextualization weights
        #contextl_hidden_states = self.gpt2_model(input_ids, position_ids=position_ids).last_hidden_state # (bs, s, d)
        contextl_hidden_states = self.gpt2_model(input_ids=input_ids, position_ids=position_ids, **kwargs).last_hidden_state
        contextualization = self.sense_weight_net(contextl_hidden_states) # (bs, nv, s, s)

        # Compute resulting outputs
        hidden_states = torch.sum(contextualization @ senses, dim=1) # (bs, nv, s, d) -> (bs, s, d)
        return BackpackGPT2BaseModelOutput(
            hidden_states=hidden_states,
            contextualization=contextualization,
            senses=senses
        )
    
    def run_with_custom_contextualization(self, input_ids, contextualization):
        # Compute senses
        sense_input_embeds = self.word_embeddings(input_ids)
        senses = self.sense_network(sense_input_embeds) # (bs, nv, s, d)

        # Compute resulting outputs
        hidden_states = torch.sum(contextualization @ senses, dim=1) # (bs, nv, s, d) -> (bs, s, d)
        return BackpackGPT2BaseModelOutput(
            hidden_states=hidden_states,
            contextualization=contextualization,
            senses=senses
        )

@dataclass
class BackpackGPT2LMHeadModelOutput(ModelOutput):
    logits: torch.FloatTensor = None
    contextualization: torch.FloatTensor = None
    backpack_hidden_states: torch.FloatTensor = None
    loss: Optional[torch.Tensor] = None                 # smoothed (for optimization)
    loss_unsmoothed: Optional[torch.Tensor] = None      # NEW: raw CE (for logging/PPL)
    senses: torch.FloatTensor = None

class BackpackGPT2LMHeadModel(BackpackGPT2PreTrainedModel, GenerationMixin):
  _keys_to_ignore_on_load_missing = [r".*attn.masked_bias", r".*attn.bias"]
  accepts_loss_kwargs = False

  def __init__(self, config):
    super().__init__(config)
    self.backpack = BackpackGPT2Model(config)
    self.lm_head = nn.Linear(config.n_embd, config.vocab_size, bias=False)

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

    self.tie_weights()

  def tie_weights(self):
      self.lm_head.weight = self.backpack.word_embeddings.weight # also tied with the underlying underlying transf

  def get_lm_head(self):
      return self.lm_head

  def get_input_embeddings(self):
    return self.backpack.word_embeddings

  def forward(
    self, 
    input_ids, 
    position_ids=None,
    labels: Optional[torch.LongTensor] = None,
    label_smoothing: Optional[float] = 0,
    cache_position: Optional[torch.LongTensor] = None,
    past_key_values: Optional[Cache] = None,
    inputs_embeds: Optional[torch.FloatTensor] = None,
    attention_mask: Optional[torch.FloatTensor] = None,
    use_cache: Optional[bool] = None,
    return_dict: Optional[bool] = None,
    **kwargs):
      outputs = self.backpack(input_ids, position_ids=position_ids, **kwargs)
      hidden_states, contextualization = outputs.hidden_states, outputs.contextualization
      senses = outputs.senses
      lm_logits = self.lm_head(hidden_states) # (bs, s, V)

      loss = None
      loss_unsmoothed = None

      if labels is not None:
        labels = labels.to(lm_logits.device)
        shift_logits = lm_logits[..., :-1, :].contiguous()
        shift_labels = labels[..., 1:].contiguous()
        
        loss_fct = CrossEntropyLoss(ignore_index=-100, reduction='mean', label_smoothing=label_smoothing)
        loss = loss_fct(shift_logits.view(-1, shift_logits.size(-1)), shift_labels.view(-1))

        # Reporting loss: **unsmoothed** (no grad) in case we used label smoothing
        with torch.no_grad():
            ce_raw = CrossEntropyLoss(ignore_index=-100, reduction="mean")
            loss_unsmoothed = ce_raw(
                shift_logits.detach().view(-1, shift_logits.size(-1)),
                shift_labels.view(-1)
            )

      return BackpackGPT2LMHeadModelOutput(
            logits=lm_logits,
            contextualization=contextualization,
            backpack_hidden_states=hidden_states,
            loss=loss,
            loss_unsmoothed=loss_unsmoothed,
            senses=senses
        )

  def run_with_custom_contextualization(self, input_ids, contextualization):
      outputs = self.backpack.run_with_custom_contextualization(input_ids, contextualization)
      hidden_states, contextualization = outputs.hidden_states, outputs.contextualization
      lm_logits = self.lm_head(hidden_states)
      return BackpackGPT2LMHeadModelOutput(
        logits=lm_logits,
        contextualization=contextualization,
        backpack_hidden_states=hidden_states,
    )