Upload trainer.json with huggingface_hub
Browse files- trainer.json +32 -0
trainer.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"repo_id": "habanoz/haber-gpt-3-40M-8k-fwb2-v1.003",
|
| 3 |
+
"trainer_config": {
|
| 4 |
+
"seed": 145,
|
| 5 |
+
"seq_length": 1024,
|
| 6 |
+
"gradient_accumulation_steps": 6,
|
| 7 |
+
"batch_size": 40,
|
| 8 |
+
"data_dir": "fineweb2tr1.003",
|
| 9 |
+
"max_iters": 20000,
|
| 10 |
+
"warmup_iters": 2000,
|
| 11 |
+
"grad_norm_clip": 1.0,
|
| 12 |
+
"out_dir": "haber-gpt-3-40M-8k-fwb2-v1.003",
|
| 13 |
+
"dtype": "float16",
|
| 14 |
+
"compile": true,
|
| 15 |
+
"gc": false,
|
| 16 |
+
"learning_rate": 0.0018,
|
| 17 |
+
"decay_lr": true,
|
| 18 |
+
"lr_decay_iters": 20000,
|
| 19 |
+
"min_lr": 0.00018,
|
| 20 |
+
"weight_decay": 0.1,
|
| 21 |
+
"beta1": 0.9,
|
| 22 |
+
"beta2": 0.95,
|
| 23 |
+
"log_interval": 50,
|
| 24 |
+
"eval_interval": 1000,
|
| 25 |
+
"eval_iters": 100,
|
| 26 |
+
"promised_flops": 65000000000000.0,
|
| 27 |
+
"wandb_log": true,
|
| 28 |
+
"wandb_project": "Haber-GPT-3-40M",
|
| 29 |
+
"wandb_run_name": "haber-gpt-3-40M-8k-fwb2-v1.003",
|
| 30 |
+
"wandb_run_id": "1733809398"
|
| 31 |
+
}
|
| 32 |
+
}
|