16 lines
366 B
YAML
16 lines
366 B
YAML
__module__: megatron.energon
|
|
__class__: Metadataset
|
|
splits:
|
|
train:
|
|
datasets:
|
|
- weight: 1.
|
|
path: <path_to_pretraining_dataset_in_energon_format>
|
|
subflavors:
|
|
augmentation: false
|
|
val:
|
|
datasets:
|
|
- weight: 1.
|
|
path: <path_to_pretraining_dataset_in_energon_format>
|
|
subflavors:
|
|
augmentation: false
|