david/flash-attention

Mirror von https://github.com/Dao-AILab/flash-attention

Autor	SHA1 Nachricht	Datum
Tri Dao	3f7d5786ba Pass alibi slopes to flash_attn_with_kvcache during generation	vor 11 Monaten
Tri Dao	2c7d7b7396 Implement norm head for Baichuan2	vor 1 Jahr
Tri Dao	c3b2196652 Add Alibi to MHA, test with Baichuan-13B	vor 1 Jahr
GAOXinyu	a8c35b4f57 FEAT: add codes which supporting for baichuan-inc/Baichuan-7B (#425)	vor 1 Jahr