 Beto
|
1f5b202c18
Adding tests for the save_metrics param in the train function
|
1 年之前 |
|
Beto
|
7474514fe0
Merging with main
|
1 年之前 |
 gaopengzhi
|
c7d410725b
Merge branch 'main' into grad_clip
|
1 年之前 |
|
gaopengzhi
|
e2797abe9b
Add gradient_clipping and gradient_clipping_threshold parameters
|
1 年之前 |
|
gaopengzhi
|
bb7c6c1e33
Support FSDP scenario
|
1 年之前 |
|
gaopengzhi
|
b1d9efd155
Refactor gradient clipping feature
|
1 年之前 |
|
Beto
|
17d02c3b44
Adding config to conditionally save stats
|
1 年之前 |
|
Beto
|
b974c87035
Merging latest from main
|
1 年之前 |
 Jeremy Howard
|
eca8410b32
Use bf16 parameters in bf16 mixed prec
|
1 年之前 |
|
gaopengzhi
|
04befdef69
Add gradient clipping feature
|
1 年之前 |
 Matthias Reso
|
e8bb7fbabc
Merge remote-tracking branch 'origin/main' into feature/length_based_batch_sampling
|
1 年之前 |
|
Matthias Reso
|
33925f71e6
Add missing amp context if use_fp16 is enabled
|
1 年之前 |
|
Beto
|
92be45b0fe
Adding matplotlib to requirements. Removing import from train_utils
|
1 年之前 |
|
Beto
|
80a6ac41b5
Adding the rank to the metrics filename to differentiate the run from each GPU
|
1 年之前 |
|
Beto
|
2f7256918e
Adding support to save fine tuning metrics to json. Adding script to plot saved metrics.
|
1 年之前 |
|
Matthias Reso
|
10f9367e56
fix missing labels in datasets
|
1 年之前 |
 hongbo.mo
|
6217635e87
Fix tqdm bar not change length after terminal is resized
|
1 年之前 |
|
Matthias Reso
|
c33ea3cacb
Fix pbar update
|
1 年之前 |
|
hongbo.mo
|
5e910e6a42
Fix typo
|
1 年之前 |
|
hongbo.mo
|
0bc6a07a80
bugfix: update tqdm bar with the fixed gradient_accumulation_steps
|
1 年之前 |
|
Matthias Reso
|
72a9832571
Merge branch 'main' into feature/package_distribution
|
1 年之前 |
|
Matthias Reso
|
ce9501f22c
remove relative imports
|
1 年之前 |
|
Matthias Reso
|
5b58afc754
Fix div by zero if run_validation=False
|
1 年之前 |
|
Matthias Reso
|
cf678b9bf0
Adjust imports to package structure + cleaned up imports
|
1 年之前 |
|
Matthias Reso
|
4c9cc7d223
Move modules into separate src folder
|
1 年之前 |
