File size: 10,854 Bytes
7ff6b5f | 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 | """
Compositional Chess Tokenizer for the Chess Challenge.
This tokenizer decomposes chess moves into meaningful components:
- Color (W/B), Piece (P/N/B/R/Q/K), Squares, Actions, Modifiers
Reduces vocabulary from 3803 to 86 tokens while enabling better generalization.
Example:
WPe2e4 -> [W, P, e2, ->, e4]
BNg8f6(x) -> [B, N, g8, x, f6]
"""
from __future__ import annotations
import json
import os
import re
from typing import Dict, List, Optional
from transformers import PreTrainedTokenizer
class ChessTokenizer(PreTrainedTokenizer):
"""Compositional tokenizer for chess moves (86 tokens vs 3803 baseline)."""
model_input_names = ["input_ids", "attention_mask"]
vocab_files_names = {"vocab_file": "vocab.json"}
PAD_TOKEN = "[PAD]"
BOS_TOKEN = "[BOS]"
EOS_TOKEN = "[EOS]"
UNK_TOKEN = "[UNK]"
# Use ASCII-safe tokens to avoid encoding issues and mismatches
MOVE_ARROW = "->"
CAPTURE_CROSS = "x"
def __init__(self, vocab_file: Optional[str] = None, vocab: Optional[Dict[str, int]] = None, **kwargs):
"""Initialize compositional chess tokenizer."""
self._pad_token = self.PAD_TOKEN
self._bos_token = self.BOS_TOKEN
self._eos_token = self.EOS_TOKEN
self._unk_token = self.UNK_TOKEN
kwargs.pop("pad_token", None)
kwargs.pop("bos_token", None)
kwargs.pop("eos_token", None)
kwargs.pop("unk_token", None)
if vocab is not None:
self._vocab = vocab
elif vocab_file is not None and os.path.exists(vocab_file):
with open(vocab_file, "r", encoding="utf-8") as f:
self._vocab = json.load(f)
else:
print("Building compositional vocabulary (86 tokens)...")
self._vocab = self._build_compositional_vocab()
self._ids_to_tokens = {v: k for k, v in self._vocab.items()}
super().__init__(
pad_token=self._pad_token,
bos_token=self._bos_token,
eos_token=self._eos_token,
unk_token=self._unk_token,
**kwargs,
)
def _build_compositional_vocab(self) -> Dict[str, int]:
"""Build 86-token compositional vocabulary."""
vocab = {}
idx = 0
# Special tokens (4 tokens)
for token in [self.PAD_TOKEN, self.BOS_TOKEN, self.EOS_TOKEN, self.UNK_TOKEN]:
vocab[token] = idx
idx += 1
# Colors (2 tokens: W, B)
for color in ["W", "B"]:
vocab[color] = idx
idx += 1
# Pieces (6 tokens: P, N, B, R, Q, K)
for piece in ["P", "N", "B", "R", "Q", "K"]:
vocab[piece] = idx
idx += 1
# Squares (64 tokens: a1-h8)
for f in ["a", "b", "c", "d", "e", "f", "g", "h"]:
for r in ["1", "2", "3", "4", "5", "6", "7", "8"]:
vocab[f + r] = idx
idx += 1
# Actions (2 tokens: →move, ×capture)
vocab[self.MOVE_ARROW] = idx
idx += 1
vocab[self.CAPTURE_CROSS] = idx
idx += 1
# Modifiers (6 tokens: +check, +*checkmate, =Q/R/B/N promotions)
for mod in ["+", "+*", "=Q", "=R", "=B", "=N"]:
vocab[mod] = idx
idx += 1
# Special moves (2 tokens: O-O, O-O-O)
for move in ["O-O", "O-O-O"]:
vocab[move] = idx
idx += 1
return vocab
@property
def vocab_size(self) -> int:
"""Return vocabulary size."""
return len(self._vocab)
def get_vocab(self) -> Dict[str, int]:
"""Return vocabulary dictionary."""
return dict(self._vocab)
def _decompose_move(self, move: str) -> List[str]:
"""Decompose chess move into component tokens.
Args:
move: Chess move in format WPe2e4 or BNg8f6(x)
Returns:
List of component tokens [color, piece, from_sq, action, to_sq, modifiers...]
"""
move = move.strip()
# Handle castling
if "O-O-O" in move or "o-o-o" in move.lower():
return [move[0], "O-O-O"] # Color + castling
elif "O-O" in move or "o-o" in move.lower():
return [move[0], "O-O"] # Color + castling
# Basic validation
if len(move) < 6:
return [self.UNK_TOKEN]
# Extract basic components
tokens = [
move[0], # Color (W/B)
move[1], # Piece (P/N/B/R/Q/K)
move[2:4] # From square (e.g., e2)
]
# Determine action (capture vs move)
is_capture = "(x)" in move or "(x+" in move or "(x+*)" in move
tokens.append(self.CAPTURE_CROSS if is_capture else self.MOVE_ARROW)
# To square
tokens.append(move[4:6])
# Add modifiers (check, checkmate)
if "(+*)" in move or "(x+*)" in move:
tokens.append("+*")
elif "(+)" in move or "(x+)" in move:
tokens.append("+")
# Handle promotions (e.g., (Q), (+Q), (xQ))
promotion_match = re.search(r'\((?:x\s*)?(?:\+\s*)?([QRBN])\)', move)
if promotion_match:
tokens.append(f"={promotion_match.group(1)}")
return tokens
def _tokenize(self, text: str) -> List[str]:
"""Tokenize string of moves into component tokens.
Args:
text: String of space-separated chess moves
Returns:
List of all component tokens
"""
all_tokens = []
for move in text.strip().split():
all_tokens.extend(self._decompose_move(move))
return all_tokens
def _convert_token_to_id(self, token: str) -> int:
"""Convert token to ID."""
return self._vocab.get(token, self._vocab.get(self.UNK_TOKEN, 3))
def _convert_id_to_token(self, index: int) -> str:
"""Convert ID to token."""
return self._ids_to_tokens.get(index, self.UNK_TOKEN)
def convert_tokens_to_string(self, tokens: List[str]) -> str:
"""Convert component tokens back to move strings.
Args:
tokens: List of component tokens
Returns:
String of reconstructed chess moves
"""
# Filter out special tokens
special = {self.PAD_TOKEN, self.BOS_TOKEN, self.EOS_TOKEN, self.UNK_TOKEN}
tokens = [t for t in tokens if t not in special]
moves = []
i = 0
while i < len(tokens):
# Handle castling
if i + 1 < len(tokens) and tokens[i + 1] in ["O-O", "O-O-O"]:
# Castling is just color + castle notation
i += 2
continue
# Regular move: need at least 5 tokens [color, piece, from_sq, action, to_sq]
if i + 4 < len(tokens):
color, piece, from_sq, action, to_sq = tokens[i:i+5]
move = f"{color}{piece}{from_sq}{to_sq}"
# Collect modifiers that follow
j = i + 5
while j < len(tokens) and tokens[j] not in ["W", "B"]:
mod = tokens[j]
if mod == "+":
move += "(+)"
elif mod == "+*":
move += "(+*)"
elif mod.startswith("="):
move += f"({mod[1]})" # Promotion
j += 1
# Add capture marker if action was capture
if action == self.CAPTURE_CROSS and "(x)" not in move:
move += "(x)"
moves.append(move)
i = j
else:
i += 1
return " ".join(moves)
def build_inputs_with_special_tokens(self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None) -> List[int]:
"""Build model inputs with BOS/EOS tokens."""
bos_id = self._vocab[self.BOS_TOKEN]
eos_id = self._vocab[self.EOS_TOKEN]
def wrap(ids: List[int]) -> List[int]:
out = list(ids)
# Only add BOS if missing
if len(out) == 0 or out[0] != bos_id:
out = [bos_id] + out
# Only add EOS if missing
if len(out) == 0 or out[-1] != eos_id:
out = out + [eos_id]
return out
if token_ids_1 is None:
return wrap(token_ids_0)
# For pair inputs, keep it simple/consistent: wrap concatenation
return wrap(token_ids_0 + token_ids_1)
def get_special_tokens_mask(
self,
token_ids_0: List[int],
token_ids_1: Optional[List[int]] = None,
already_has_special_tokens: bool = False
) -> List[int]:
"""Return mask identifying special tokens."""
bos_id = self._vocab[self.BOS_TOKEN]
eos_id = self._vocab[self.EOS_TOKEN]
pad_id = self._vocab[self.PAD_TOKEN]
unk_id = self._vocab[self.UNK_TOKEN]
def mask_for(ids: List[int]) -> List[int]:
return [1 if tid in {bos_id, eos_id, pad_id, unk_id} else 0 for tid in ids]
if already_has_special_tokens:
if token_ids_1 is None:
return mask_for(token_ids_0)
return mask_for(token_ids_0 + token_ids_1)
# No special tokens yet: build mask consistent with build_inputs_with_special_tokens
if token_ids_1 is None:
return [1] + ([0] * len(token_ids_0)) + [1]
return [1] + ([0] * len(token_ids_0)) + ([0] * len(token_ids_1)) + [1]
def create_token_type_ids_from_sequences(
self,
token_ids_0: List[int],
token_ids_1: Optional[List[int]] = None
) -> List[int]:
"""Create token type IDs (all zeros for single sequence type)."""
# Build final ids consistently with build_inputs_with_special_tokens
final_ids = self.build_inputs_with_special_tokens(token_ids_0, token_ids_1)
return [0] * len(final_ids)
def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] = None) -> tuple:
"""Save vocabulary to JSON file."""
if not os.path.isdir(save_directory):
os.makedirs(save_directory, exist_ok=True)
vocab_file = os.path.join(
save_directory,
(filename_prefix + "-" if filename_prefix else "") + "vocab.json"
)
with open(vocab_file, "w", encoding="utf-8") as f:
json.dump(self._vocab, f, ensure_ascii=False, indent=2)
return (vocab_file,)
|