Commit
•
3fc9006
1
Parent(s):
ad8be43
Feat(deepspeed): Add zero2 config (#476)
Browse files* zero2 config
* config added
* linting
---------
Co-authored-by: mhenrichsen <some_email@hey.com>
- deepspeed/zero2.json +46 -0
deepspeed/zero2.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"zero_optimization": {
|
| 3 |
+
"stage": 2,
|
| 4 |
+
"offload_optimizer": {
|
| 5 |
+
"device": "cpu"
|
| 6 |
+
},
|
| 7 |
+
"contiguous_gradients": true,
|
| 8 |
+
"overlap_comm": true
|
| 9 |
+
},
|
| 10 |
+
"bf16": {
|
| 11 |
+
"enabled": "auto"
|
| 12 |
+
},
|
| 13 |
+
"fp16": {
|
| 14 |
+
"enabled": "auto",
|
| 15 |
+
"auto_cast": false,
|
| 16 |
+
"loss_scale": 0,
|
| 17 |
+
"initial_scale_power": 32,
|
| 18 |
+
"loss_scale_window": 1000,
|
| 19 |
+
"hysteresis": 2,
|
| 20 |
+
"min_loss_scale": 1
|
| 21 |
+
},
|
| 22 |
+
"optimizer": {
|
| 23 |
+
"type": "AdamW",
|
| 24 |
+
"params": {
|
| 25 |
+
"lr": "auto",
|
| 26 |
+
"betas": [
|
| 27 |
+
0.9,
|
| 28 |
+
0.999
|
| 29 |
+
],
|
| 30 |
+
"eps": 1e-8,
|
| 31 |
+
"weight_decay": "auto"
|
| 32 |
+
}
|
| 33 |
+
},
|
| 34 |
+
"scheduler": {
|
| 35 |
+
"type": "WarmupDecayLR",
|
| 36 |
+
"params": {
|
| 37 |
+
"warmup_min_lr": "auto",
|
| 38 |
+
"warmup_max_lr": "auto",
|
| 39 |
+
"warmup_num_steps": "auto",
|
| 40 |
+
"total_num_steps": "auto"
|
| 41 |
+
}
|
| 42 |
+
},
|
| 43 |
+
"train_batch_size": "auto",
|
| 44 |
+
"train_micro_batch_size_per_gpu": "auto",
|
| 45 |
+
"wall_clock_breakdown": false
|
| 46 |
+
}
|