modeling_bvv_max.py

from transformers import PreTrainedModel, PretrainedConfig
from transformers.modeling_outputs import CausalLMOutputWithCrossAttentions
import torch
import torch.nn as nn
from torch.nn import functional as F
from transformers.modeling_outputs import CausalLMOutput

class BVVMaxConfig(PretrainedConfig):
    model_type = "bvv_max"
    
    def __init__(
        self,
        vocab_size = 131072,
        n_embd = 1024,
        n_head = 12,  
        n_layer = 12, 
        block_size = 1024,
        pad_id = 57344,
        **kwargs
    ):
        super().__init__(**kwargs)
        self.vocab_size = vocab_size
        self.block_size  = block_size 
        self.n_embd = n_embd
        self.n_layer = n_layer
        self.n_head = n_head
        self.pad_id = pad_id 

class SimpleSelfAttentionHead(nn.Module):
    def __init__(self, head_size, n_embd, block_size):
        super().__init__()
        self.q_proj = nn.Linear(n_embd, head_size, bias=False)
        self.k_proj = nn.Linear(n_embd, head_size, bias=False)
        self.v_proj = nn.Linear(n_embd, head_size, bias=False)
        self.o_proj = nn.Linear(head_size, head_size, bias=False)

        self.dropout = nn.Dropout(0.0)
        self.register_buffer('tril', torch.tril(torch.ones(block_size, block_size)))

    def forward(self, x):
        B, T, C = x.shape
        q = self.q_proj(x)     # (B,T,head_size)
        k = self.k_proj(x)
        v = self.v_proj(x)

        attn_scores = q @ k.transpose(-2, -1) * (k.shape[-1] ** -0.5)  # (B,T,T)
        
        attn_scores = attn_scores.masked_fill(self.tril[:T, :T] == 0, torch.finfo(attn_scores.dtype).min) #float('-inf'))

        attn_probs = F.softmax(attn_scores, dim=-1)
        attn_probs = self.dropout(attn_probs)

        out = attn_probs @ v   # (B,T,head_size)
        out = self.o_proj(out) # (B,T,head_size)

        return out

class SimpleMultiHeadSelfAttention(nn.Module):
    def __init__(self, n_embd, n_head, block_size):
        super().__init__()
        self.head_size = n_embd // n_head
        self.heads = nn.ModuleList([SimpleSelfAttentionHead(self.head_size, n_embd, block_size) for _ in range(n_head)])
        self.out_proj = nn.Linear(n_head * self.head_size, n_embd)
        self.dropout = nn.Dropout(0.0)

    def forward(self, x):
        out = torch.cat([head(x) for head in self.heads], dim=-1) 
        out = self.dropout(self.out_proj(out))
        return out

class TransformerMLP(nn.Module):
    def __init__(self, n_embd):
        super().__init__()
        self.net = nn.Sequential(
            nn.Linear(n_embd, 4 * n_embd),
            nn.GELU(),
            nn.Linear(4 * n_embd, n_embd),
            nn.Dropout(0.0),
        )

    def forward(self, x):
        return self.net(x)

class TransformerBlock(nn.Module):
    def __init__(self, n_embd, n_head, block_size):
        super().__init__()
        self.self_attn = SimpleMultiHeadSelfAttention(n_embd, n_head, block_size)
        self.mlp = TransformerMLP(n_embd)
        self.input_layernorm = nn.LayerNorm(n_embd)
        self.post_attention_layernorm = nn.LayerNorm(n_embd)

    def forward(self, x):
        x = x + self.self_attn(self.input_layernorm(x))
        x = x + self.mlp(self.post_attention_layernorm(x))
        return x

class BVVMaxForCausalLM(PreTrainedModel):
    config_class = BVVMaxConfig
    
    def __init__(self, config):
        super().__init__(config)
        self.token_embeddings = nn.Embedding(config.vocab_size, config.n_embd)
        self.position_embeddings = nn.Embedding(config.block_size, config.n_embd)
        self.transformer_layers = nn.Sequential(*[
            TransformerBlock(config.n_embd, n_head=config.n_head, block_size=config.block_size) for _ in range(config.n_layer)
        ])
        self.final_layernorm = nn.LayerNorm(config.n_embd)
        self.lm_head = nn.Linear(config.n_embd, config.vocab_size)

        self.apply(self._init_weights)

    def _init_weights(self, module):
        if isinstance(module, nn.Linear):
            torch.nn.init.normal_(module.weight, mean=0.0, std=0.02)
            if module.bias is not None:
                torch.nn.init.zeros_(module.bias)
        elif isinstance(module, nn.Embedding):
            torch.nn.init.normal_(module.weight, mean=0.0, std=0.02)


    def forward(self, idx, targets=None):
        B, T = idx.shape

        positions = torch.arange(0, T, device=idx.device).unsqueeze(0).expand(B, T)

        x = self.token_embeddings(idx) + self.position_embeddings(positions)
        
        x = self.transformer_layers(x)
        x = self.final_layernorm(x)
        logits = self.lm_head(x)

        loss = None
        if targets is not None:
            #logits_flat = logits.view(-1, logits.size(-1))
            #targets_flat = targets.view(-1)
            logits_flat = logits.reshape(-1, logits.size(-1))
            targets_flat = targets.reshape(-1)
            loss = F.cross_entropy(logits_flat, targets_flat, ignore_index = 57344)

        return CausalLMOutput(
            logits=logits,
            loss=loss,
        )

    def generate(self, 
                input_ids=None,
                max_new_tokens=None,
                max_length=None,
                temperature=1.0,
                top_k=None,
                top_p=None,
                do_sample=True,
                pad_token_id=None,
                eos_token_id=None,
                **kwargs):
        
        if input_ids is None:
            raise ValueError("Input_ids must be provided")
        
        idx = input_ids
        
        if max_new_tokens is None:
            if max_length is not None:
                max_new_tokens = max_length - idx.shape[1]
            else:
                max_new_tokens = 50  
        
        with torch.no_grad():
            for _ in range(max_new_tokens):
                idx_cond = idx[:, -self.config.block_size:]
                
                outputs = self(idx_cond)
                logits = outputs.logits[:, -1, :] / temperature
                
                if top_k is not None:
                    v, _ = torch.topk(logits, min(top_k, logits.size(-1)))
                    logits[logits < v[:, [-1]]] = float('-inf')
                
                if top_p is not None:
                    sorted_logits, sorted_indices = torch.sort(logits, descending=True)
                    cumulative_probs = torch.cumsum(F.softmax(sorted_logits, dim=-1), dim=-1)
                    
                    sorted_indices_to_remove = cumulative_probs > top_p
                    sorted_indices_to_remove[..., 1:] = sorted_indices_to_remove[..., :-1].clone()
                    sorted_indices_to_remove[..., 0] = 0
                    
                    indices_to_remove = sorted_indices_to_remove.scatter(1, sorted_indices, sorted_indices_to_remove)
                    logits[indices_to_remove] = float('-inf')
                
                probs = F.softmax(logits, dim=-1)
                
                if do_sample:
                    idx_next = torch.multinomial(probs, num_samples=1)
                else:
                    idx_next = torch.argmax(logits, dim=-1, keepdim=True)
                
                idx = torch.cat((idx, idx_next), dim=1)
                
    
                if eos_token_id is not None and (idx_next == eos_token_id).any():
                    break
        
        return idx