sophia_opt.SophiaG
Browse files
scripts/pretrain-model.yaml
CHANGED
|
@@ -65,12 +65,10 @@ train:
|
|
| 65 |
log_interval: 1
|
| 66 |
|
| 67 |
# Number of samples between optimizer steps across data-parallel ranks (type: int, default: 512)
|
| 68 |
-
|
| 69 |
-
global_batch_size: 256
|
| 70 |
|
| 71 |
# Number of samples per data-parallel rank (type: int, default: 4)
|
| 72 |
-
|
| 73 |
-
micro_batch_size: 4
|
| 74 |
|
| 75 |
# Number of iterations with learning rate warmup active (type: int, default: 2000)
|
| 76 |
lr_warmup_steps: 2000
|
|
|
|
| 65 |
log_interval: 1
|
| 66 |
|
| 67 |
# Number of samples between optimizer steps across data-parallel ranks (type: int, default: 512)
|
| 68 |
+
global_batch_size: 512
|
|
|
|
| 69 |
|
| 70 |
# Number of samples per data-parallel rank (type: int, default: 4)
|
| 71 |
+
micro_batch_size: 2
|
|
|
|
| 72 |
|
| 73 |
# Number of iterations with learning rate warmup active (type: int, default: 2000)
|
| 74 |
lr_warmup_steps: 2000
|