pretrain base 0
Browse files
scripts/pretrain_base_model_0.yaml
CHANGED
|
@@ -61,7 +61,7 @@ train:
|
|
| 61 |
global_batch_size: 512
|
| 62 |
|
| 63 |
# Number of samples per data-parallel rank (type: int, default: 4)
|
| 64 |
-
micro_batch_size:
|
| 65 |
|
| 66 |
# Number of iterations with learning rate warmup active (type: int, default: 2000)
|
| 67 |
lr_warmup_steps: 100
|
|
|
|
| 61 |
global_batch_size: 512
|
| 62 |
|
| 63 |
# Number of samples per data-parallel rank (type: int, default: 4)
|
| 64 |
+
micro_batch_size: 2
|
| 65 |
|
| 66 |
# Number of iterations with learning rate warmup active (type: int, default: 2000)
|
| 67 |
lr_warmup_steps: 100
|