 Tri Dao
|
2c996ca25f
Use SeqlenInfo for bwd and epilogue
|
1 týždeň pred |
|
Tri Dao
|
29cdfedd80
Use Bulk reduce instead of TMA for dQaccum, split across WGs
|
1 týždeň pred |
|
Tri Dao
|
82dc825759
Don't use the unsafe convert_type function
|
2 týždňov pred |
|
Tri Dao
|
6e8b25e426
Refactor
|
2 mesiacov pred |
 Ying Zhang
|
db80387343
Add seqused_q in fwd / bwd and seqused_k in bwd.
|
3 mesiacov pred |
 Cameron Shinn
|
3cea2fb6ee
Add ArchTag to pre/postprocess bwd kernels (#1180)
|
3 mesiacov pred |
|
Tri Dao
|
bafe253042
[FA3] Bwd
|
4 mesiacov pred |
