Skip to content

[Bugfix] Use reshape_and_cache for num_kv_heads > 1 in KunlunAttentionImpl #263

[Bugfix] Use reshape_and_cache for num_kv_heads > 1 in KunlunAttentionImpl

[Bugfix] Use reshape_and_cache for num_kv_heads > 1 in KunlunAttentionImpl #263