engdarwish commited on
Commit
3c22c21
·
verified ·
1 Parent(s): 767fe26

Fix broken GitHub link, add Quick Start + inference.py usage docs

Browse files
Files changed (1) hide show
  1. README.md +38 -8
README.md CHANGED
@@ -3,9 +3,9 @@ license: mit
3
  tags: [chess, pytorch, resnet, se-resnet, position-evaluation, policy-head, arabic]
4
  language: [ar, en]
5
  ---
6
- # ♟️ Chess Position Evaluator V7 HULK
7
 
8
- > **Ahmed Darwish** | [@engdarwish](https://huggingface.co/engdarwish)
9
 
10
  ## Architecture
11
  SE-ResNet-20 with Dual Head (Value + Policy)
@@ -15,14 +15,44 @@ SE-ResNet-20 with Dual Head (Value + Policy)
15
  | Parameters | 47,450,753 |
16
  | Best Val Loss | 0.030174 |
17
  | Architecture | SE-ResNet-20 |
18
- | Input | 18×8×8 board tensor |
19
  | Value Output | [-1, 1] (tanh) |
20
- | Policy Output | 20,480 move logits |
21
 
22
  ## Training Data
23
- - **Kaggle Chess Evaluations** — 12.9M positions (Stockfish evals)
24
- - **Lichess 2023-10** 2M moves from ELO 2000+ games
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
25
 
26
  ## Contact
27
- 📧 eahmeddarwish@gmail.com
28
- 🌐 [GitHub](https://github.com/engrdarwish)
 
3
  tags: [chess, pytorch, resnet, se-resnet, position-evaluation, policy-head, arabic]
4
  language: [ar, en]
5
  ---
6
+ # Chess Position Evaluator - V7 HULK
7
 
8
+ Ahmed Darwish | [@engdarwish](https://huggingface.co/engdarwish)
9
 
10
  ## Architecture
11
  SE-ResNet-20 with Dual Head (Value + Policy)
 
15
  | Parameters | 47,450,753 |
16
  | Best Val Loss | 0.030174 |
17
  | Architecture | SE-ResNet-20 |
18
+ | Input | 18x8x8 board tensor |
19
  | Value Output | [-1, 1] (tanh) |
20
+ | Policy Output | 20,480 move logits (from-square x to-square x underpromotion) |
21
 
22
  ## Training Data
23
+
24
+ | Source | Size |
25
+ |--------|------|
26
+ | Kaggle Chess Evaluations | 12.9M positions (Stockfish evals) |
27
+ | Lichess 2023-10 | 2M moves from ELO 2000+ games |
28
+
29
+ ## Quick Start
30
+
31
+ ```bash
32
+ pip install torch python-chess huggingface_hub numpy
33
+ ```
34
+
35
+ ```python
36
+ from inference import load_model, get_best_move
37
+ import chess
38
+
39
+ model = load_model()
40
+ value, moves = get_best_move(model, chess.STARTING_FEN, top_k=3)
41
+ print(value, moves)
42
+ ```
43
+
44
+ See inference.py in this repo for the full, runnable example: board encoding, model definition, and legal-move-masked move ranking.
45
+
46
+ ## Files
47
+
48
+ | File | Purpose |
49
+ |------|---------|
50
+ | model_weights.pt (190MB) | Inference weights: use this to load the model |
51
+ | checkpoint.pt (569MB) | Full training checkpoint (model + optimizer + scheduler state): only needed to resume training |
52
+ | config.json | Architecture config |
53
+ | inference.py | Minimal working inference example |
54
+ | training_curves.png | Training/validation loss curves |
55
 
56
  ## Contact
57
+ eahmeddarwish@gmail.com
58
+ [GitHub](https://github.com/eahmeddarwish)