Chess Challenge submission by alexandreduplessis

README.md

@@ -0,0 +1,26 @@
+---
+library_name: transformers
+tags:
+- chess
+- llm-course
+- chess-challenge
+license: mit
+---
+
+# chesstoktok
+
+Chess model submitted to the LLM Course Chess Challenge.
+
+## Submission Info
+
+- **Submitted by**: [alexandreduplessis](https://huggingface.co/alexandreduplessis)
+- **Parameters**: 703,872
+- **Organization**: LLM-course
+
+## Model Details
+
+- **Architecture**: Chess Transformer (GPT-style)
+- **Vocab size**: 73
+- **Embedding dim**: 128
+- **Layers**: 4
+- **Heads**: 4

config.json

@@ -0,0 +1,20 @@
+{
+  "architectures": [
+    "ChessForCausalLM"
+  ],
+  "bos_token_id": 1,
+  "dropout": 0.05,
+  "eos_token_id": 2,
+  "layer_norm_epsilon": 1e-05,
+  "model_type": "chess_transformer",
+  "n_ctx": 256,
+  "n_embd": 128,
+  "n_head": 4,
+  "n_inner": 384,
+  "n_layer": 4,
+  "pad_token_id": 0,
+  "tie_weights": true,
+  "torch_dtype": "float32",
+  "transformers_version": "4.55.0",
+  "vocab_size": 73
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:375a8dd55b7164613a088c79e5dcf56fcd0a753f4abb07b7056f9a518af71d9d
+size 2819888

special_tokens_map.json

@@ -0,0 +1,6 @@
+{
+  "bos_token": "[BOS]",
+  "eos_token": "[EOS]",
+  "pad_token": "[PAD]",
+  "unk_token": "[UNK]"
+}

tokenizer.py

@@ -0,0 +1,194 @@
+from __future__ import annotations
+
+import json
+import os
+import re
+from typing import Dict, List, Optional, Tuple
+
+from transformers import PreTrainedTokenizer
+
+
+class ChessTokenizer(PreTrainedTokenizer):
+
+    model_input_names = ["input_ids", "attention_mask"]
+    vocab_files_names = {"vocab_file": "vocab.json"}
+
+    # Special tokens
+    PAD_TOKEN = "[PAD]"
+    BOS_TOKEN = "[BOS]"
+    EOS_TOKEN = "[EOS]"
+    UNK_TOKEN = "[UNK]"
+
+    # Structure token
+    MOVE_TOKEN = "[MOVE]"
+
+    _MOVE_RE = re.compile(
+        r'^(?P<color>[WB])(?P<piece>[PNBRQK])(?P<from>[a-h][1-8])(?P<to>[a-h][1-8])(?P<rest>.*)$'
+    )
+    _PROMO_RE = re.compile(r'=?([QRBNqrbn])')
+
+    def __init__(
+        self,
+        vocab_file: Optional[str] = None,
+        vocab: Optional[Dict[str, int]] = None,
+        **kwargs,
+    ):
+        # Initialize special tokens
+        self._pad_token = self.PAD_TOKEN
+        self._bos_token = self.BOS_TOKEN
+        self._eos_token = self.EOS_TOKEN
+        self._unk_token = self.UNK_TOKEN
+
+        # Remove any duplicate special-token entries passed through kwargs
+        kwargs.pop("pad_token", None)
+        kwargs.pop("bos_token", None)
+        kwargs.pop("eos_token", None)
+        kwargs.pop("unk_token", None)
+
+        # Load or create vocabulary
+        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:
+            self._vocab = self._create_default_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 _create_default_vocab(self) -> Dict[str, int]:
+        special = [self.PAD_TOKEN, self.BOS_TOKEN, self.EOS_TOKEN, self.UNK_TOKEN, self.MOVE_TOKEN]
+        return {t: i for i, t in enumerate(special)}
+
+    @classmethod
+    def build_structured_vocab(cls) -> "ChessTokenizer":
+        special = [cls.PAD_TOKEN, cls.BOS_TOKEN, cls.EOS_TOKEN, cls.UNK_TOKEN, cls.MOVE_TOKEN]
+
+        files = "abcdefgh"
+        ranks = "12345678"
+        squares = [f"{f}{r}" for f in files for r in ranks]  # 64
+
+        promo = [f"promo_{p}" for p in ("q", "r", "b", "n")]
+
+        tokens = special + squares + promo
+        vocab = {t: i for i, t in enumerate(tokens)}
+        return cls(vocab=vocab)
+
+    @classmethod
+    def build_vocab_from_dataset(
+        cls,
+        dataset_name: str = "dlouapre/lichess_2025-01_1M",
+        split: str = "train",
+        column: str = "text",
+        min_frequency: int = 500,
+        max_samples: Optional[int] = 100000,
+    ) -> "ChessTokenizer":
+        return cls.build_structured_vocab()
+
+    @property
+    def vocab_size(self) -> int:
+        return len(self._vocab)
+
+    def get_vocab(self) -> Dict[str, int]:
+        return dict(self._vocab)
+
+    def _convert_token_to_id(self, token: str) -> int:
+        return self._vocab.get(token, self._vocab.get(self.UNK_TOKEN, 0))
+
+    def _convert_id_to_token(self, index: int) -> str:
+        return self._ids_to_tokens.get(index, self.UNK_TOKEN)
+
+    def convert_tokens_to_string(self, tokens: List[str]) -> str:
+        drop = {self.PAD_TOKEN, self.BOS_TOKEN, self.EOS_TOKEN, self.UNK_TOKEN}
+        return " ".join(t for t in tokens if t not in drop)
+
+    def _decompose_one_move(self, move_tok: str) -> List[str]:
+        m = self._MOVE_RE.match(move_tok)
+        if not m:
+            return [self.UNK_TOKEN]
+
+        from_sq = m.group("from")
+        to_sq = m.group("to")
+        rest = m.group("rest") or ""
+
+        out = [self.MOVE_TOKEN, from_sq, to_sq]
+
+        # Promotion detection (best-effort)
+        pm = self._PROMO_RE.search(rest)
+        if pm:
+            p = pm.group(1).lower()
+            if p in ("q", "r", "b", "n"):
+                out.append(f"promo_{p}")
+
+        return out
+
+    def _tokenize(self, text: str) -> List[str]:
+        text = text.strip()
+        if not text:
+            return []
+
+        special = {self.PAD_TOKEN, self.BOS_TOKEN, self.EOS_TOKEN, self.UNK_TOKEN, self.MOVE_TOKEN}
+
+        if " " not in text:
+            if text in special:
+                return [text]
+            if text in self._vocab:
+                return [text]
+            return self._decompose_one_move(text)
+
+        out: List[str] = []
+        for part in text.split():
+            if part in special:
+                out.append(part)
+            elif part in self._vocab:
+                out.append(part)
+            else:
+                out.extend(self._decompose_one_move(part))
+        return out
+
+    def save_vocabulary(
+        self,
+        save_directory: str,
+        filename_prefix: Optional[str] = None,
+    ) -> Tuple[str]:
+        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,)
+
+
+def count_vocab_from_dataset(
+    dataset_name: str = "dlouapre/lichess_2025-01_1M",
+    split: str = "train",
+    column: str = "text",
+    max_samples: Optional[int] = 10000,
+) -> Dict[str, int]:
+    from collections import Counter
+    from datasets import load_dataset
+
+    dataset = load_dataset(dataset_name, split=split)
+    if max_samples is not None:
+        dataset = dataset.select(range(min(max_samples, len(dataset))))
+
+    token_counts = Counter()
+    for example in dataset:
+        moves = example[column].strip().split()
+        token_counts.update(moves)
+
+    return dict(token_counts)

tokenizer_config.json

@@ -0,0 +1,50 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "[PAD]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "[BOS]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "[EOS]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "[UNK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "auto_map": {
+    "AutoTokenizer": [
+      "tokenizer.ChessTokenizer",
+      null
+    ]
+  },
+  "bos_token": "[BOS]",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "[EOS]",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "[PAD]",
+  "tokenizer_class": "ChessTokenizer",
+  "unk_token": "[UNK]"
+}

vocab.json

@@ -0,0 +1,75 @@
+{
+  "[PAD]": 0,
+  "[BOS]": 1,
+  "[EOS]": 2,
+  "[UNK]": 3,
+  "[MOVE]": 4,
+  "a1": 5,
+  "a2": 6,
+  "a3": 7,
+  "a4": 8,
+  "a5": 9,
+  "a6": 10,
+  "a7": 11,
+  "a8": 12,
+  "b1": 13,
+  "b2": 14,
+  "b3": 15,
+  "b4": 16,
+  "b5": 17,
+  "b6": 18,
+  "b7": 19,
+  "b8": 20,
+  "c1": 21,
+  "c2": 22,
+  "c3": 23,
+  "c4": 24,
+  "c5": 25,
+  "c6": 26,
+  "c7": 27,
+  "c8": 28,
+  "d1": 29,
+  "d2": 30,
+  "d3": 31,
+  "d4": 32,
+  "d5": 33,
+  "d6": 34,
+  "d7": 35,
+  "d8": 36,
+  "e1": 37,
+  "e2": 38,
+  "e3": 39,
+  "e4": 40,
+  "e5": 41,
+  "e6": 42,
+  "e7": 43,
+  "e8": 44,
+  "f1": 45,
+  "f2": 46,
+  "f3": 47,
+  "f4": 48,
+  "f5": 49,
+  "f6": 50,
+  "f7": 51,
+  "f8": 52,
+  "g1": 53,
+  "g2": 54,
+  "g3": 55,
+  "g4": 56,
+  "g5": 57,
+  "g6": 58,
+  "g7": 59,
+  "g8": 60,
+  "h1": 61,
+  "h2": 62,
+  "h3": 63,
+  "h4": 64,
+  "h5": 65,
+  "h6": 66,
+  "h7": 67,
+  "h8": 68,
+  "promo_q": 69,
+  "promo_r": 70,
+  "promo_b": 71,
+  "promo_n": 72
+}