Update README.md on pretraining_dataset (#360)
Browse files* Update README.md on pretraining_dataset
* Fix message
README.md
CHANGED
|
@@ -505,6 +505,9 @@ torchdistx_path:
|
|
| 505 |
# Set padding for data collator to 'longest'
|
| 506 |
collator_pad_to_longest:
|
| 507 |
|
|
|
|
|
|
|
|
|
|
| 508 |
# Debug mode
|
| 509 |
debug:
|
| 510 |
|
|
|
|
| 505 |
# Set padding for data collator to 'longest'
|
| 506 |
collator_pad_to_longest:
|
| 507 |
|
| 508 |
+
# Set to HF dataset for type: 'completion' for streaming instead of pre-tokenize
|
| 509 |
+
pretraining_dataset:
|
| 510 |
+
|
| 511 |
# Debug mode
|
| 512 |
debug:
|
| 513 |
|