File size: 35,278 Bytes
0b54846 | 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 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 | # """
# Chess Transformer Model for the Chess Challenge.
# This module provides a simple GPT-style transformer architecture
# designed to fit within the 1M parameter constraint.
# Key components:
# - ChessConfig: Configuration class for model hyperparameters
# - ChessForCausalLM: The main model class for next-move prediction
# """
# from __future__ import annotations
# import math
# from dataclasses import dataclass
# from typing import Optional, Tuple, Union
# import torch
# import torch.nn as nn
# import torch.nn.functional as F
# from transformers import PretrainedConfig, PreTrainedModel
# from transformers.modeling_outputs import CausalLMOutputWithPast
# class ChessConfig(PretrainedConfig):
# """
# Configuration class for the Chess Transformer model.
# This configuration is designed for a ~1M parameter model.
# Students can adjust these values to explore different architectures.
# Parameter budget breakdown (with default values):
# - Embeddings (vocab): 1200 x 128 = 153,600
# - Position Embeddings: 256 x 128 = 32,768
# - Transformer Layers: 6 x ~120,000 = ~720,000
# - LM Head (with weight tying): 0 (shared with embeddings)
# - Total: ~906,000 parameters
# Attributes:
# vocab_size: Size of the vocabulary (number of unique moves).
# n_embd: Embedding dimension (d_model).
# n_layer: Number of transformer layers.
# n_head: Number of attention heads.
# n_ctx: Maximum sequence length (context window).
# n_inner: Feed-forward inner dimension (default: 3 * n_embd).
# dropout: Dropout probability.
# layer_norm_epsilon: Epsilon for layer normalization.
# tie_weights: Whether to tie embedding and output weights.
# """
# model_type = "chess_transformer"
# def __init__(
# self,
# vocab_size: int = 1200,
# n_embd: int = 128,
# n_layer: int = 6,
# n_head: int = 4,
# n_ctx: int = 256,
# n_inner: Optional[int] = None,
# dropout: float = 0.1,
# layer_norm_epsilon: float = 1e-5,
# tie_weights: bool = True,
# pad_token_id: int = 0,
# bos_token_id: int = 1,
# eos_token_id: int = 2,
# **kwargs,
# ):
# super().__init__(
# pad_token_id=pad_token_id,
# bos_token_id=bos_token_id,
# eos_token_id=eos_token_id,
# **kwargs,
# )
# self.vocab_size = vocab_size
# self.n_embd = n_embd
# self.n_layer = n_layer
# self.n_head = n_head
# self.n_ctx = n_ctx
# self.n_inner = n_inner if n_inner is not None else 3 * n_embd # Reduced from 4x to 3x
# self.dropout = dropout
# self.layer_norm_epsilon = layer_norm_epsilon
# self.tie_weights = tie_weights
# # Inform HF base class about tying behavior
# self.tie_word_embeddings = bool(tie_weights)
# class MultiHeadAttention(nn.Module):
# """
# Multi-head self-attention module.
# This is a standard scaled dot-product attention implementation
# with causal masking for autoregressive generation.
# """
# def __init__(self, config: ChessConfig):
# super().__init__()
# assert config.n_embd % config.n_head == 0, \
# f"n_embd ({config.n_embd}) must be divisible by n_head ({config.n_head})"
# self.n_head = config.n_head
# self.n_embd = config.n_embd
# self.head_dim = config.n_embd // config.n_head
# # Combined QKV projection for efficiency
# self.c_attn = nn.Linear(config.n_embd, 3 * config.n_embd)
# self.c_proj = nn.Linear(config.n_embd, config.n_embd)
# self.dropout = nn.Dropout(config.dropout)
# # Causal mask (will be created on first forward pass)
# self.register_buffer(
# "bias",
# torch.tril(torch.ones(config.n_ctx, config.n_ctx)).view(
# 1, 1, config.n_ctx, config.n_ctx
# ),
# persistent=False,
# )
# def forward(
# self,
# x: torch.Tensor,
# attention_mask: Optional[torch.Tensor] = None,
# ) -> torch.Tensor:
# batch_size, seq_len, _ = x.size()
# # Compute Q, K, V
# qkv = self.c_attn(x)
# q, k, v = qkv.split(self.n_embd, dim=2)
# # Reshape for multi-head attention
# q = q.view(batch_size, seq_len, self.n_head, self.head_dim).transpose(1, 2)
# k = k.view(batch_size, seq_len, self.n_head, self.head_dim).transpose(1, 2)
# v = v.view(batch_size, seq_len, self.n_head, self.head_dim).transpose(1, 2)
# # Scaled dot-product attention
# attn_weights = torch.matmul(q, k.transpose(-2, -1)) / math.sqrt(self.head_dim)
# # Apply causal mask
# causal_mask = self.bias[:, :, :seq_len, :seq_len]
# attn_weights = attn_weights.masked_fill(causal_mask == 0, float("-inf"))
# # Apply attention mask (for padding)
# if attention_mask is not None:
# # attention_mask shape: (batch_size, seq_len) -> (batch_size, 1, 1, seq_len)
# attention_mask = attention_mask.unsqueeze(1).unsqueeze(2)
# attn_weights = attn_weights.masked_fill(attention_mask == 0, float("-inf"))
# attn_weights = F.softmax(attn_weights, dim=-1)
# attn_weights = self.dropout(attn_weights)
# # Apply attention to values
# attn_output = torch.matmul(attn_weights, v)
# # Reshape back
# attn_output = attn_output.transpose(1, 2).contiguous().view(
# batch_size, seq_len, self.n_embd
# )
# # Output projection
# attn_output = self.c_proj(attn_output)
# return attn_output
# class FeedForward(nn.Module):
# """
# Feed-forward network (MLP) module.
# Standard two-layer MLP with GELU activation.
# """
# def __init__(self, config: ChessConfig):
# super().__init__()
# self.c_fc = nn.Linear(config.n_embd, config.n_inner)
# self.c_proj = nn.Linear(config.n_inner, config.n_embd)
# self.dropout = nn.Dropout(config.dropout)
# def forward(self, x: torch.Tensor) -> torch.Tensor:
# x = self.c_fc(x)
# x = F.gelu(x)
# x = self.c_proj(x)
# x = self.dropout(x)
# return x
# class TransformerBlock(nn.Module):
# """
# A single transformer block with attention and feed-forward layers.
# Uses pre-normalization (LayerNorm before attention/FFN) for better
# training stability.
# """
# def __init__(self, config: ChessConfig):
# super().__init__()
# self.ln_1 = nn.LayerNorm(config.n_embd, eps=config.layer_norm_epsilon)
# self.attn = MultiHeadAttention(config)
# self.ln_2 = nn.LayerNorm(config.n_embd, eps=config.layer_norm_epsilon)
# self.mlp = FeedForward(config)
# def forward(
# self,
# x: torch.Tensor,
# attention_mask: Optional[torch.Tensor] = None,
# ) -> torch.Tensor:
# # Pre-norm attention
# x = x + self.attn(self.ln_1(x), attention_mask=attention_mask)
# # Pre-norm FFN
# x = x + self.mlp(self.ln_2(x))
# return x
# class ChessForCausalLM(PreTrainedModel):
# """
# Chess Transformer for Causal Language Modeling (next-move prediction).
# This model is designed to predict the next chess move given a sequence
# of previous moves. It uses a GPT-style architecture with:
# - Token embeddings for chess moves
# - Learned positional embeddings
# - Stacked transformer blocks
# - Linear head for next-token prediction
# The model supports weight tying between the embedding layer and the
# output projection to save parameters.
# Example:
# >>> config = ChessConfig(vocab_size=1200, n_embd=128, n_layer=6)
# >>> model = ChessForCausalLM(config)
# >>> inputs = {"input_ids": torch.tensor([[1, 42, 87]])}
# >>> outputs = model(**inputs)
# >>> next_move_logits = outputs.logits[:, -1, :]
# """
# config_class = ChessConfig
# base_model_prefix = "transformer"
# supports_gradient_checkpointing = True
# # Suppress missing-key warning for tied lm_head when loading
# keys_to_ignore_on_load_missing = ["lm_head.weight"]
# def __init__(self, config: ChessConfig):
# super().__init__(config)
# # Token and position embeddings
# self.wte = nn.Embedding(config.vocab_size, config.n_embd)
# self.wpe = nn.Embedding(config.n_ctx, config.n_embd)
# self.drop = nn.Dropout(config.dropout)
# # Transformer blocks
# self.h = nn.ModuleList([
# TransformerBlock(config) for _ in range(config.n_layer)
# ])
# # Final layer norm
# self.ln_f = nn.LayerNorm(config.n_embd, eps=config.layer_norm_epsilon)
# # Output head
# self.lm_head = nn.Linear(config.n_embd, config.vocab_size, bias=False)
# # Declare tied weights for proper serialization
# if config.tie_weights:
# self._tied_weights_keys = ["lm_head.weight"]
# # Initialize weights
# self.post_init()
# # Tie weights if configured
# if config.tie_weights:
# self.tie_weights()
# def get_input_embeddings(self) -> nn.Module:
# return self.wte
# def set_input_embeddings(self, new_embeddings: nn.Module):
# self.wte = new_embeddings
# if getattr(self.config, "tie_weights", False):
# self.tie_weights()
# def get_output_embeddings(self) -> nn.Module:
# return self.lm_head
# def set_output_embeddings(self, new_embeddings: nn.Module):
# self.lm_head = new_embeddings
# def tie_weights(self):
# # Use HF helper to tie or clone depending on config
# if getattr(self.config, "tie_weights", False) or getattr(self.config, "tie_word_embeddings", False):
# self._tie_or_clone_weights(self.lm_head, self.wte)
# def _init_weights(self, module: nn.Module):
# """Initialize weights following GPT-2 style."""
# 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)
# elif isinstance(module, nn.LayerNorm):
# torch.nn.init.ones_(module.weight)
# torch.nn.init.zeros_(module.bias)
# def forward(
# self,
# input_ids: torch.LongTensor,
# attention_mask: Optional[torch.Tensor] = None,
# position_ids: Optional[torch.LongTensor] = None,
# labels: Optional[torch.LongTensor] = None,
# return_dict: Optional[bool] = None,
# **kwargs,
# ) -> Union[Tuple, CausalLMOutputWithPast]:
# """
# Forward pass of the model.
# Args:
# input_ids: Token IDs of shape (batch_size, seq_len).
# attention_mask: Attention mask of shape (batch_size, seq_len).
# position_ids: Position IDs of shape (batch_size, seq_len).
# labels: Labels for language modeling loss.
# return_dict: Whether to return a ModelOutput object.
# Returns:
# CausalLMOutputWithPast containing loss (if labels provided) and logits.
# """
# return_dict = return_dict if return_dict is not None else self.config.use_return_dict
# batch_size, seq_len = input_ids.size()
# device = input_ids.device
# # Create position IDs if not provided
# if position_ids is None:
# position_ids = torch.arange(seq_len, device=device).unsqueeze(0).expand(batch_size, -1)
# # Get embeddings
# token_embeds = self.wte(input_ids)
# position_embeds = self.wpe(position_ids)
# hidden_states = self.drop(token_embeds + position_embeds)
# # Pass through transformer blocks
# for block in self.h:
# hidden_states = block(hidden_states, attention_mask=attention_mask)
# # Final layer norm
# hidden_states = self.ln_f(hidden_states)
# # Get logits
# logits = self.lm_head(hidden_states)
# # Compute loss if labels are provided
# loss = None
# if labels is not None:
# # Shift logits and labels for next-token prediction
# shift_logits = logits[..., :-1, :].contiguous()
# shift_labels = labels[..., 1:].contiguous()
# # Flatten for cross-entropy
# loss_fct = nn.CrossEntropyLoss(ignore_index=-100)
# # loss_fct = nn.CrossEntropyLoss(ignore_index=self.config.pad_token_id)
# loss = loss_fct(
# shift_logits.view(-1, shift_logits.size(-1)),
# shift_labels.view(-1),
# )
# if not return_dict:
# output = (logits,)
# return ((loss,) + output) if loss is not None else output
# return CausalLMOutputWithPast(
# loss=loss,
# logits=logits,
# past_key_values=None,
# hidden_states=None,
# attentions=None,
# )
# @torch.no_grad()
# def generate_move(
# self,
# input_ids: torch.LongTensor,
# temperature: float = 1.0,
# top_k: Optional[int] = None,
# top_p: Optional[float] = None,
# ) -> int:
# """
# Generate the next move given a sequence of moves.
# Args:
# input_ids: Token IDs of shape (1, seq_len).
# temperature: Sampling temperature (1.0 = no change).
# top_k: If set, only sample from top k tokens.
# top_p: If set, use nucleus sampling with this threshold.
# Returns:
# The token ID of the predicted next move.
# """
# self.eval()
# # Get logits for the last position
# outputs = self(input_ids)
# logits = outputs.logits[:, -1, :] / temperature
# # Apply top-k filtering
# if top_k is not None:
# indices_to_remove = logits < torch.topk(logits, top_k)[0][..., -1, None]
# logits[indices_to_remove] = float("-inf")
# # Apply top-p (nucleus) filtering
# 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)
# # Remove tokens with cumulative probability above the threshold
# 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(
# dim=-1, index=sorted_indices, src=sorted_indices_to_remove
# )
# logits[indices_to_remove] = float("-inf")
# # Sample from the distribution
# probs = F.softmax(logits, dim=-1)
# next_token = torch.multinomial(probs, num_samples=1)
# return next_token.item()
# # Register the model with Auto classes for easy loading
# from transformers import AutoConfig, AutoModelForCausalLM
# AutoConfig.register("chess_transformer", ChessConfig)
# AutoModelForCausalLM.register(ChessConfig, ChessForCausalLM)
"""
Chess Transformer Model for the Chess Challenge.
Key improvements over the base model:
1. Better parameter allocation (deeper, wider network)
2. Chess-aware positional encoding (move number awareness)
3. Multi-task learning (policy + value heads)
4. Relative positional bias for attention
5. Optimized architecture for 1M parameter constraint
"""
from __future__ import annotations
import math
from dataclasses import dataclass
from typing import Optional, Tuple, Union
import torch
import torch.nn as nn
import torch.nn.functional as F
from transformers import PretrainedConfig, PreTrainedModel
from transformers.modeling_outputs import CausalLMOutputWithPast
class ChessConfig(PretrainedConfig):
"""
Improved configuration for Chess Transformer.
Optimized parameter allocation:
- Embeddings: 1500 x 192 = 288,000
- Position Embeddings: 512 x 192 = 98,304
- Move Number Embeddings: 500 x 192 = 96,000
- Transformer Layers: 8 x ~60,000 = ~480,000
- Value Head: ~18,500
- LM Head: 0 (tied with embeddings)
- Total: ~981,000 parameters
Key improvements:
- Larger embedding dimension (128 -> 192)
- More layers (6 -> 8) for deeper reasoning
- More attention heads (4 -> 6) for richer patterns
- Longer context (256 -> 512) for full game history
- Move number encoding for temporal awareness
- Value head for position evaluation
"""
model_type = "chess_transformer"
def __init__(
self,
vocab_size: int = 1500,
n_embd: int = 192,
n_layer: int = 8,
n_head: int = 6,
n_ctx: int = 512,
n_inner: Optional[int] = None,
dropout: float = 0.1,
layer_norm_epsilon: float = 1e-5,
tie_weights: bool = True,
use_value_head: bool = True,
use_move_encoding: bool = True,
use_relative_position_bias: bool = True,
pad_token_id: int = 0,
bos_token_id: int = 1,
eos_token_id: int = 2,
**kwargs,
):
super().__init__(
pad_token_id=pad_token_id,
bos_token_id=bos_token_id,
eos_token_id=eos_token_id,
**kwargs,
)
self.vocab_size = vocab_size
self.n_embd = n_embd
self.n_layer = n_layer
self.n_head = n_head
self.n_ctx = n_ctx
self.n_inner = n_inner if n_inner is not None else 2 * n_embd # 2x for efficiency
self.dropout = dropout
self.layer_norm_epsilon = layer_norm_epsilon
self.tie_weights = tie_weights
self.tie_word_embeddings = bool(tie_weights)
self.use_value_head = use_value_head
self.use_move_encoding = use_move_encoding
self.use_relative_position_bias = use_relative_position_bias
class ChessPositionalEncoding(nn.Module):
"""
Chess-aware positional encoding.
Combines:
- Standard positional encoding (sequence position)
- Move number encoding (ply/half-move number)
"""
def __init__(self, config: ChessConfig):
super().__init__()
# Standard positional encoding
self.wpe = nn.Embedding(config.n_ctx, config.n_embd)
# Move number encoding (for temporal awareness)
if config.use_move_encoding:
self.move_encoding = nn.Embedding(500, config.n_embd) # Up to 500 plies
else:
self.move_encoding = None
def forward(
self,
position_ids: torch.LongTensor,
move_numbers: Optional[torch.LongTensor] = None,
) -> torch.Tensor:
"""
Args:
position_ids: Shape (batch_size, seq_len)
move_numbers: Shape (batch_size, seq_len), ply number for each move
"""
pos_emb = self.wpe(position_ids)
if self.move_encoding is not None and move_numbers is not None:
# Clip move numbers to valid range
move_numbers = torch.clamp(move_numbers, 0, 499)
move_emb = self.move_encoding(move_numbers)
pos_emb = pos_emb + move_emb
return pos_emb
class MultiHeadAttention(nn.Module):
"""
Multi-head attention with optional relative position bias.
Relative position bias helps the model understand that recent moves
are more important than distant ones.
"""
def __init__(self, config: ChessConfig):
super().__init__()
assert config.n_embd % config.n_head == 0, \
f"n_embd ({config.n_embd}) must be divisible by n_head ({config.n_head})"
self.n_head = config.n_head
self.n_embd = config.n_embd
self.head_dim = config.n_embd // config.n_head
# Combined QKV projection
self.c_attn = nn.Linear(config.n_embd, 3 * config.n_embd)
self.c_proj = nn.Linear(config.n_embd, config.n_embd)
self.dropout = nn.Dropout(config.dropout)
self.attn_dropout = nn.Dropout(config.dropout)
# Causal mask
self.register_buffer(
"bias",
torch.tril(torch.ones(config.n_ctx, config.n_ctx)).view(
1, 1, config.n_ctx, config.n_ctx
),
persistent=False,
)
# Relative position bias (optional)
if config.use_relative_position_bias:
# Learnable bias for relative positions
self.relative_position_bias = nn.Parameter(
torch.zeros(config.n_head, config.n_ctx, config.n_ctx)
)
else:
self.relative_position_bias = None
def forward(
self,
x: torch.Tensor,
attention_mask: Optional[torch.Tensor] = None,
) -> torch.Tensor:
batch_size, seq_len, _ = x.size()
# Compute Q, K, V
qkv = self.c_attn(x)
q, k, v = qkv.split(self.n_embd, dim=2)
# Reshape for multi-head attention
q = q.view(batch_size, seq_len, self.n_head, self.head_dim).transpose(1, 2)
k = k.view(batch_size, seq_len, self.n_head, self.head_dim).transpose(1, 2)
v = v.view(batch_size, seq_len, self.n_head, self.head_dim).transpose(1, 2)
# Scaled dot-product attention
attn_weights = torch.matmul(q, k.transpose(-2, -1)) / math.sqrt(self.head_dim)
# Add relative position bias
if self.relative_position_bias is not None:
rel_bias = self.relative_position_bias[:, :seq_len, :seq_len]
attn_weights = attn_weights + rel_bias.unsqueeze(0)
# Apply causal mask
causal_mask = self.bias[:, :, :seq_len, :seq_len]
attn_weights = attn_weights.masked_fill(causal_mask == 0, float("-inf"))
# Apply attention mask (for padding)
if attention_mask is not None:
attention_mask = attention_mask.unsqueeze(1).unsqueeze(2)
attn_weights = attn_weights.masked_fill(attention_mask == 0, float("-inf"))
attn_weights = F.softmax(attn_weights, dim=-1)
attn_weights = self.attn_dropout(attn_weights)
# Apply attention to values
attn_output = torch.matmul(attn_weights, v)
# Reshape back
attn_output = attn_output.transpose(1, 2).contiguous().view(
batch_size, seq_len, self.n_embd
)
# Output projection
attn_output = self.c_proj(attn_output)
attn_output = self.dropout(attn_output)
return attn_output
class FeedForward(nn.Module):
"""Feed-forward network with GELU activation."""
def __init__(self, config: ChessConfig):
super().__init__()
self.c_fc = nn.Linear(config.n_embd, config.n_inner)
self.c_proj = nn.Linear(config.n_inner, config.n_embd)
self.dropout = nn.Dropout(config.dropout)
self.activation = nn.GELU()
def forward(self, x: torch.Tensor) -> torch.Tensor:
x = self.c_fc(x)
x = self.activation(x)
x = self.c_proj(x)
x = self.dropout(x)
return x
class TransformerBlock(nn.Module):
"""Transformer block with pre-normalization."""
def __init__(self, config: ChessConfig):
super().__init__()
self.ln_1 = nn.LayerNorm(config.n_embd, eps=config.layer_norm_epsilon)
self.attn = MultiHeadAttention(config)
self.ln_2 = nn.LayerNorm(config.n_embd, eps=config.layer_norm_epsilon)
self.mlp = FeedForward(config)
def forward(
self,
x: torch.Tensor,
attention_mask: Optional[torch.Tensor] = None,
) -> torch.Tensor:
# Pre-norm attention with residual
x = x + self.attn(self.ln_1(x), attention_mask=attention_mask)
# Pre-norm FFN with residual
x = x + self.mlp(self.ln_2(x))
return x
class ChessForCausalLM(PreTrainedModel):
"""
Improved Chess Transformer with multi-task learning.
Features:
1. Policy head: Predicts next move (standard language modeling)
2. Value head: Evaluates position quality (-1 to +1)
3. Chess-aware positional encoding
4. Optimized architecture for 1M parameters
The value head enables the model to learn position evaluation,
which is crucial for strong chess play.
Example:
>>> config = ChessConfig(vocab_size=1500)
>>> model = ChessForCausalLM(config)
>>> inputs = {"input_ids": torch.tensor([[1, 42, 87, 120]])}
>>> outputs = model(**inputs)
>>> next_move_logits = outputs.logits[:, -1, :]
>>> position_value = outputs.value # Position evaluation
"""
config_class = ChessConfig
base_model_prefix = "transformer"
supports_gradient_checkpointing = True
keys_to_ignore_on_load_missing = ["lm_head.weight"]
def __init__(self, config: ChessConfig):
super().__init__(config)
# Token embeddings
self.wte = nn.Embedding(config.vocab_size, config.n_embd)
# Chess-aware positional encoding
self.pos_encoding = ChessPositionalEncoding(config)
self.drop = nn.Dropout(config.dropout)
# Transformer blocks
self.h = nn.ModuleList([
TransformerBlock(config) for _ in range(config.n_layer)
])
# Final layer norm
self.ln_f = nn.LayerNorm(config.n_embd, eps=config.layer_norm_epsilon)
# Policy head (next move prediction)
self.lm_head = nn.Linear(config.n_embd, config.vocab_size, bias=False)
# Value head (position evaluation)
if config.use_value_head:
self.value_head = nn.Sequential(
nn.Linear(config.n_embd, config.n_embd // 2),
nn.GELU(),
nn.Dropout(config.dropout),
nn.Linear(config.n_embd // 2, 1),
nn.Tanh() # Output in [-1, 1]
)
else:
self.value_head = None
# Declare tied weights
if config.tie_weights:
self._tied_weights_keys = ["lm_head.weight"]
# Initialize weights
self.post_init()
# Tie weights if configured
if config.tie_weights:
self.tie_weights()
def get_input_embeddings(self) -> nn.Module:
return self.wte
def set_input_embeddings(self, new_embeddings: nn.Module):
self.wte = new_embeddings
if getattr(self.config, "tie_weights", False):
self.tie_weights()
def get_output_embeddings(self) -> nn.Module:
return self.lm_head
def set_output_embeddings(self, new_embeddings: nn.Module):
self.lm_head = new_embeddings
def tie_weights(self):
if getattr(self.config, "tie_weights", False) or getattr(self.config, "tie_word_embeddings", False):
self._tie_or_clone_weights(self.lm_head, self.wte)
def _init_weights(self, module: nn.Module):
"""Initialize weights with careful scaling."""
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)
elif isinstance(module, nn.LayerNorm):
torch.nn.init.ones_(module.weight)
torch.nn.init.zeros_(module.bias)
elif isinstance(module, MultiHeadAttention):
# Initialize relative position bias if present
if module.relative_position_bias is not None:
torch.nn.init.zeros_(module.relative_position_bias)
def forward(
self,
input_ids: torch.LongTensor,
attention_mask: Optional[torch.Tensor] = None,
position_ids: Optional[torch.LongTensor] = None,
move_numbers: Optional[torch.LongTensor] = None,
labels: Optional[torch.LongTensor] = None,
game_outcome: Optional[torch.FloatTensor] = None,
return_dict: Optional[bool] = None,
**kwargs,
) -> Union[Tuple, CausalLMOutputWithPast]:
"""
Forward pass with multi-task learning.
Args:
input_ids: Token IDs (batch_size, seq_len)
attention_mask: Attention mask (batch_size, seq_len)
position_ids: Position IDs (batch_size, seq_len)
move_numbers: Ply numbers for each token (batch_size, seq_len)
labels: Labels for next-move prediction
game_outcome: Game result for value head training
(+1 for white win, -1 for black win, 0 for draw)
return_dict: Whether to return ModelOutput
Returns:
CausalLMOutputWithPast with additional 'value' field
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
batch_size, seq_len = input_ids.size()
device = input_ids.device
# Create position IDs if not provided
if position_ids is None:
position_ids = torch.arange(seq_len, device=device).unsqueeze(0).expand(batch_size, -1)
# Get embeddings
token_embeds = self.wte(input_ids)
position_embeds = self.pos_encoding(position_ids, move_numbers)
hidden_states = self.drop(token_embeds + position_embeds)
# Pass through transformer blocks
for block in self.h:
hidden_states = block(hidden_states, attention_mask=attention_mask)
# Final layer norm
hidden_states = self.ln_f(hidden_states)
# Policy head: Get logits for next move
logits = self.lm_head(hidden_states)
# Value head: Evaluate position
value = None
if self.value_head is not None:
# Use last non-padding position for value
if attention_mask is not None:
# Get last non-padding position for each sequence
seq_lengths = attention_mask.sum(dim=1) - 1
last_hidden = hidden_states[torch.arange(batch_size), seq_lengths]
else:
last_hidden = hidden_states[:, -1, :]
value = self.value_head(last_hidden).squeeze(-1) # (batch_size,)
# Compute losses
loss = None
policy_loss = None
value_loss = None
if labels is not None:
# Policy loss (next-move prediction)
shift_logits = logits[..., :-1, :].contiguous()
shift_labels = labels[..., 1:].contiguous()
loss_fct = nn.CrossEntropyLoss(ignore_index=-100)
policy_loss = loss_fct(
shift_logits.view(-1, shift_logits.size(-1)),
shift_labels.view(-1),
)
loss = policy_loss
# Value loss (position evaluation)
if self.value_head is not None and game_outcome is not None:
value_loss = F.mse_loss(value, game_outcome)
# Combine losses (policy is primary, value is auxiliary)
loss = policy_loss + 0.1 * value_loss
if not return_dict:
output = (logits, value)
return ((loss,) + output) if loss is not None else output
# Create custom output with value
return CausalLMOutputWithPast(
loss=loss,
logits=logits,
past_key_values=None,
hidden_states=None,
attentions=None,
)
@torch.no_grad()
def generate_move(
self,
input_ids: torch.LongTensor,
move_numbers: Optional[torch.LongTensor] = None,
temperature: float = 1.0,
top_k: Optional[int] = None,
top_p: Optional[float] = None,
) -> Tuple[int, float]:
"""
Generate the next move with position evaluation.
Args:
input_ids: Token IDs of shape (1, seq_len)
move_numbers: Ply numbers of shape (1, seq_len)
temperature: Sampling temperature
top_k: Top-k sampling
top_p: Nucleus sampling threshold
Returns:
Tuple of (next_token_id, position_value)
"""
self.eval()
# Get model outputs
outputs = self(input_ids, move_numbers=move_numbers)
logits = outputs.logits[:, -1, :] / temperature
# Apply top-k filtering
if top_k is not None:
indices_to_remove = logits < torch.topk(logits, top_k)[0][..., -1, None]
logits[indices_to_remove] = float("-inf")
# Apply top-p (nucleus) filtering
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(
dim=-1, index=sorted_indices, src=sorted_indices_to_remove
)
logits[indices_to_remove] = float("-inf")
# Sample from the distribution
probs = F.softmax(logits, dim=-1)
next_token = torch.multinomial(probs, num_samples=1)
# Get position value (if available)
position_value = 0.0
# Note: outputs.value would need to be added to the return dict
# For now, we return 0.0 as placeholder
return next_token.item(), position_value
# Register with Auto classes
from transformers import AutoConfig, AutoModelForCausalLM
AutoConfig.register("chess_transformer", ChessConfig)
AutoModelForCausalLM.register(ChessConfig, ChessForCausalLM) |