 Hamid Shojanazeri
|
d51d2cce9c
adding sdpa for flash attn
|
hai 1 ano |
|
Hamid Shojanazeri
|
9a2434f408
update the fine-tuning script
|
hai 1 ano |
|
Hamid Shojanazeri
|
162be4c045
Revert "Flop counter, profiling and GC (#357)"
|
hai 1 ano |
|
Hamid Shojanazeri
|
71d137c722
Merge branch 'main' into flop_counter_gc
|
hai 1 ano |
 Less Wright
|
3f2c33e4f8
Update finetuning.py - remove nightly check
|
hai 1 ano |
|
Hamid Shojanazeri
|
b15ffeeaf4
clean up
|
hai 1 ano |
|
Hamid Shojanazeri
|
19089269d3
add gc
|
hai 1 ano |
 Abhilash Majumder
|
4793f0fdf3
Merge branch 'main' into ipex_feature
|
hai 1 ano |
|
Abhilash Majumder
|
6a78b96764
Merge branch 'main' into ipex_feature
|
hai 1 ano |
 Matthias Reso
|
4c225c65eb
Fix order of concat vs sampler
|
hai 1 ano |
|
Matthias Reso
|
5da84b2913
Fix usage of dataclass for train_config and fsdp_config
|
hai 1 ano |
|
Matthias Reso
|
653a79e3dd
Invalidate context in labels for samsum + grammar
|
hai 1 ano |
|
Matthias Reso
|
a647955fc8
Make packing/padding a training setting
|
hai 1 ano |
|
Matthias Reso
|
2e4bd2a665
Resize vocab size to fix idx error
|
hai 1 ano |
|
Matthias Reso
|
ca41c1c697
Adjust tests to len based batch sampling
|
hai 1 ano |
 Shijie Wu
|
91e2573aa8
pass weight_decay into optimizer
|
hai 1 ano |
 Howard Liberty
|
cc356b6017
Add FSDP CPU offloading option
|
hai 1 ano |
 abhilash1910
|
ad6b27d316
merge conflicts
|
hai 1 ano |
|
abhilash1910
|
33da341af5
upstream resolve conflict
|
hai 1 ano |
|
Matthias Reso
|
eb69e754a0
Merge branch 'main' into feature/package_distribution
|
hai 1 ano |
|
Matthias Reso
|
72a9832571
Merge branch 'main' into feature/package_distribution
|
hai 1 ano |
|
Matthias Reso
|
ce9501f22c
remove relative imports
|
hai 1 ano |
|
Matthias Reso
|
5b58afc754
Fix div by zero if run_validation=False
|
hai 1 ano |
|
Matthias Reso
|
cf678b9bf0
Adjust imports to package structure + cleaned up imports
|
hai 1 ano |
|
Matthias Reso
|
4c9cc7d223
Move modules into separate src folder
|
hai 1 ano |
