jleezhang commited on
Commit
bed8516
·
verified ·
1 Parent(s): c3907ac

Epoch 10 - Training started: 2025-11-15 19:24:39

Browse files
Files changed (4) hide show
  1. README.md +6 -6
  2. checkpoint_epoch_10.pt +2 -2
  3. config.json +6 -3
  4. pytorch_model.bin +2 -2
README.md CHANGED
@@ -6,7 +6,7 @@ NNUE (Efficiently Updatable Neural Network) chess evaluation model
6
 
7
  - **Model type**: NNUE Chess Evaluation
8
  - **Architecture**: HalfKP feature representation
9
- - **Uploaded**: 2025-11-15 19:17:53
10
 
11
  ## Architecture
12
 
@@ -28,9 +28,9 @@ Output Layer: 32 → 1 (centipawn evaluation)
28
 
29
  ## Training Information
30
 
31
- - **Epoch**: 64
32
- - **Training Loss**: 2944877.7457
33
- - **Validation Loss**: 2821792.7308
34
 
35
  ## Usage
36
 
@@ -61,8 +61,8 @@ model_config = checkpoint['model_config']
61
  ## Training Configuration
62
 
63
  - **batch_size**: 256
64
- - **learning_rate**: 0.001
65
- - **num_epochs**: 100
66
  - **optimizer**: adam
67
  - **loss_function**: mse
68
  - **hidden_size**: 256
 
6
 
7
  - **Model type**: NNUE Chess Evaluation
8
  - **Architecture**: HalfKP feature representation
9
+ - **Uploaded**: 2025-11-15 19:27:29
10
 
11
  ## Architecture
12
 
 
28
 
29
  ## Training Information
30
 
31
+ - **Epoch**: 10
32
+ - **Training Loss**: 3120906.5251
33
+ - **Validation Loss**: 2923721.9199
34
 
35
  ## Usage
36
 
 
61
  ## Training Configuration
62
 
63
  - **batch_size**: 256
64
+ - **learning_rate**: 0.003
65
+ - **num_epochs**: 50
66
  - **optimizer**: adam
67
  - **loss_function**: mse
68
  - **hidden_size**: 256
checkpoint_epoch_10.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:95ecaed5ba60bcbcd9b034d120d31efc0b7fd58dcafec3361a5167195a4623a3
3
- size 126054539
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fe4f79c0dd385b53af3bb7d104ae650b21bee0e676802c5eb2fe8323c07a98f4
3
+ size 126054987
config.json CHANGED
@@ -10,9 +10,11 @@
10
  "train_data_path": "data/train.jsonl",
11
  "val_data_path": "data/val.jsonl",
12
  "batch_size": 256,
13
- "learning_rate": 0.001,
14
- "num_epochs": 100,
15
  "weight_decay": 1e-05,
 
 
16
  "hidden_size": 256,
17
  "hidden2_size": 32,
18
  "hidden3_size": 32,
@@ -29,6 +31,8 @@
29
  "scheduler": null,
30
  "scheduler_step_size": 30,
31
  "scheduler_gamma": 0.1,
 
 
32
  "loss_function": "mse",
33
  "log_every_n_batches": 100,
34
  "verbose": true,
@@ -43,7 +47,6 @@
43
  "download_positions_per_game": 10,
44
  "download_num_workers": 4,
45
  "download_batch_size": 100,
46
- "download_output_format": "jsonl",
47
  "download_rated_only": true,
48
  "download_output_dir": "data",
49
  "download_mode": "streaming",
 
10
  "train_data_path": "data/train.jsonl",
11
  "val_data_path": "data/val.jsonl",
12
  "batch_size": 256,
13
+ "learning_rate": 0.003,
14
+ "num_epochs": 50,
15
  "weight_decay": 1e-05,
16
+ "max_grad_norm": 1.0,
17
+ "early_stopping_patience": 10,
18
  "hidden_size": 256,
19
  "hidden2_size": 32,
20
  "hidden3_size": 32,
 
31
  "scheduler": null,
32
  "scheduler_step_size": 30,
33
  "scheduler_gamma": 0.1,
34
+ "warmup_epochs": 5,
35
+ "warmup_start_lr": 1e-05,
36
  "loss_function": "mse",
37
  "log_every_n_batches": 100,
38
  "verbose": true,
 
47
  "download_positions_per_game": 10,
48
  "download_num_workers": 4,
49
  "download_batch_size": 100,
 
50
  "download_rated_only": true,
51
  "download_output_dir": "data",
52
  "download_mode": "streaming",
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:cb969826c34dd87a0854234b70e304044a2af0ba8e7725a4fa2cc4fdeef7671c
3
- size 42017825
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d729af8233d872c00bd05b36bd5ee3d9b11a38b3d6d9abbeac94c7e0855d2bba
3
+ size 42017953