Upload colm_config.json with huggingface_hub
Browse files- colm_config.json +38 -0
colm_config.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architecture": {
|
| 3 |
+
"n_embd": 512,
|
| 4 |
+
"n_layer": 12,
|
| 5 |
+
"embed_dim": 512,
|
| 6 |
+
"block_size": 256,
|
| 7 |
+
"vocab_size": 499
|
| 8 |
+
},
|
| 9 |
+
"training": {
|
| 10 |
+
"batch_size": 2,
|
| 11 |
+
"max_iters": 1000000,
|
| 12 |
+
"learning_rate": 6e-4,
|
| 13 |
+
"min_lr": 6e-5,
|
| 14 |
+
"warmup_iters": 20000,
|
| 15 |
+
"weight_decay": 0.1,
|
| 16 |
+
"grad_clip": 1
|
| 17 |
+
},
|
| 18 |
+
"evaluation": {
|
| 19 |
+
"eval_interval": 25000,
|
| 20 |
+
"eval_iters": 100,
|
| 21 |
+
"save_interval": 25000,
|
| 22 |
+
"patience": 15
|
| 23 |
+
},
|
| 24 |
+
"lr_schedule": {
|
| 25 |
+
"lr_cycle_length": 25000,
|
| 26 |
+
"lr_cycle_warmup": 10000,
|
| 27 |
+
"lr_decay_rate": 0.95
|
| 28 |
+
},
|
| 29 |
+
"scanner": {
|
| 30 |
+
"scanner_clamp": 70.0
|
| 31 |
+
},
|
| 32 |
+
"paths": {
|
| 33 |
+
"checkpoint_path": "checkpoints/colm_checkpoint_big.pt",
|
| 34 |
+
"best_checkpoint_path": "checkpoints/colm_best_big.pt",
|
| 35 |
+
"tokenizer_path": "colm_tokenizer.json",
|
| 36 |
+
"dataset_path": "datasets/DCDM_big_dataset.txt"
|
| 37 |
+
}
|
| 38 |
+
}
|