Update README.md on pretraining_dataset (#360)
* Update README.md on pretraining_dataset * Fix message
This commit is contained in:
@@ -505,6 +505,9 @@ torchdistx_path:
|
|||||||
# Set padding for data collator to 'longest'
|
# Set padding for data collator to 'longest'
|
||||||
collator_pad_to_longest:
|
collator_pad_to_longest:
|
||||||
|
|
||||||
|
# Set to HF dataset for type: 'completion' for streaming instead of pre-tokenize
|
||||||
|
pretraining_dataset:
|
||||||
|
|
||||||
# Debug mode
|
# Debug mode
|
||||||
debug:
|
debug:
|
||||||
|
|
||||||
|
|||||||
Reference in New Issue
Block a user