Upload train_budget_hours_24.json with huggingface_hub
Browse files- train_budget_hours_24.json +42 -0
train_budget_hours_24.json
ADDED
|
@@ -0,0 +1,42 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"optim": {
|
| 3 |
+
"type": "AdamW",
|
| 4 |
+
"lr": 0.001,
|
| 5 |
+
"betas": [
|
| 6 |
+
0.9,
|
| 7 |
+
0.98
|
| 8 |
+
],
|
| 9 |
+
"eps": 1e-12,
|
| 10 |
+
"weight_decay": 0.01,
|
| 11 |
+
"amsgrad": false
|
| 12 |
+
},
|
| 13 |
+
"optim_mod": {
|
| 14 |
+
"name": "none"
|
| 15 |
+
},
|
| 16 |
+
"name": "bert-o3",
|
| 17 |
+
"limited_decay_keys": [
|
| 18 |
+
"bias",
|
| 19 |
+
"LayerNorm.bias",
|
| 20 |
+
"LayerNorm.weight"
|
| 21 |
+
],
|
| 22 |
+
"warmup_steps": 0,
|
| 23 |
+
"cooldown_steps": 0,
|
| 24 |
+
"steps": 600000,
|
| 25 |
+
"scheduler": "budget-triangle2",
|
| 26 |
+
"batch_size": 4096,
|
| 27 |
+
"batch_size_ramp": 300000,
|
| 28 |
+
"gradient_clipping": 0.5,
|
| 29 |
+
"pretrain_in_train_mode": false,
|
| 30 |
+
"objective": {
|
| 31 |
+
"name": "masked-lm",
|
| 32 |
+
"mlm_probability": 0.15,
|
| 33 |
+
"use_80_20_rule": true,
|
| 34 |
+
"disable_mlm": false,
|
| 35 |
+
"token_drop": 0.0
|
| 36 |
+
},
|
| 37 |
+
"reverse_dataset_order": false,
|
| 38 |
+
"budget": 24,
|
| 39 |
+
"gradinit": {
|
| 40 |
+
"enabled": false
|
| 41 |
+
}
|
| 42 |
+
}
|