* fwd var-seq-len * fixes * benchmark * fixes --------- Co-authored-by: Tri Dao <tridao@users.noreply.github.com> |
||
|---|---|---|
| .. | ||
| layers | ||
| losses | ||
| models | ||
| modules | ||
| ops | ||
| pyproject.toml | ||
| test_flash_attn.py | ||
| test_rotary.py | ||
| test_util.py | ||
* fwd var-seq-len * fixes * benchmark * fixes --------- Co-authored-by: Tri Dao <tridao@users.noreply.github.com> |
||
|---|---|---|
| .. | ||
| layers | ||
| losses | ||
| models | ||
| modules | ||
| ops | ||
| pyproject.toml | ||
| test_flash_attn.py | ||
| test_rotary.py | ||
| test_util.py | ||