File size: 3,406 Bytes
f669547
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81

import torch
import torch.nn as nn
from transformers import PreTrainedModel, PretrainedConfig, GenerationMixin
from transformers.modeling_outputs import CausalLMOutput

class ChessConfig(PretrainedConfig):
    model_type = "chess_transformer"
    def __init__(self, vocab_size=1000, n_embd=128, n_layer=4, n_head=4, n_inner=512, n_ctx=256, **kwargs):
        super().__init__(**kwargs)
        self.vocab_size = vocab_size
        self.n_embd = n_embd
        self.n_layer = n_layer
        self.n_head = n_head
        self.n_inner = n_inner
        self.n_ctx = n_ctx
        
        # Alias pour compatibilité Hugging Face (Important pour .generate)
        self.num_hidden_layers = n_layer
        self.hidden_size = n_embd
        self.num_attention_heads = n_head

class Block(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.ln1 = nn.LayerNorm(config.n_embd)
        self.attn = nn.MultiheadAttention(config.n_embd, config.n_head, batch_first=True)
        self.ln2 = nn.LayerNorm(config.n_embd)
        self.mlp = nn.Sequential(nn.Linear(config.n_embd, config.n_inner), nn.GELU(), nn.Linear(config.n_inner, config.n_embd))
        
    def forward(self, x, mask=None):
        attn_out, _ = self.attn(self.ln1(x), self.ln1(x), self.ln1(x), attn_mask=mask, need_weights=False)
        return x + attn_out + self.mlp(self.ln2(x + attn_out))

class ChessForCausalLM(PreTrainedModel, GenerationMixin):
    config_class = ChessConfig
    
    def __init__(self, config):
        super().__init__(config)
        self.config = config
        self.token_emb = nn.Embedding(config.vocab_size, config.n_embd)
        self.pos_emb = nn.Embedding(config.n_ctx, config.n_embd)
        self.blocks = nn.ModuleList([Block(config) for _ in range(config.n_layer)])
        self.ln_f = nn.LayerNorm(config.n_embd)
        self.lm_head = nn.Linear(config.n_embd, config.vocab_size, bias=False)
        
        # Weight Tying (Partage de poids)
        self.lm_head.weight = self.token_emb.weight
        self.apply(self._init_weights)
        
    def _init_weights(self, module):
        if isinstance(module, (nn.Linear, nn.Embedding)): 
            torch.nn.init.normal_(module.weight, mean=0.0, std=0.02)
            
    # Obligatoire pour .generate()
    def prepare_inputs_for_generation(self, input_ids, **kwargs):
        return {"input_ids": input_ids}
            
    def forward(self, input_ids, attention_mask=None, labels=None, **kwargs):
        B, T = input_ids.shape
        x = self.token_emb(input_ids) + self.pos_emb(torch.arange(T, device=input_ids.device))
        
        # Masque causal (Interdiction de voir le futur)
        mask = torch.triu(torch.ones(T, T, device=input_ids.device) * float('-inf'), diagonal=1)
        
        for block in self.blocks: 
            x = block(x, mask=mask)
            
        logits = self.lm_head(self.ln_f(x))
        
        loss = None
        if labels is not None:
            # Shift des labels (input [BOS, A, B] -> predit [A, B, C])
            # Grâce à l'ajout de [BOS] par le tokenizer, cette formule est correcte.
            loss = nn.CrossEntropyLoss(ignore_index=-100)(
                logits[..., :-1, :].contiguous().view(-1, self.config.vocab_size), 
                labels[..., 1:].contiguous().view(-1)
            )
            
        return CausalLMOutput(loss=loss, logits=logits)