[Kernel] unified_attention for Attention.forward (#11967)
Signed-off-by: Chen Zhang <zhangch99@outlook.com>
This commit is contained in:
@@ -2171,5 +2171,4 @@ def bind_kv_cache(
|
||||
forward_ctx = ctx[layer_name]
|
||||
assert len(forward_ctx.kv_cache) == len(kv_cache)
|
||||
for ve, ve_kv_cache in enumerate(kv_cache):
|
||||
assert forward_ctx.kv_cache[ve].numel() == 0
|
||||
forward_ctx.kv_cache[ve] = ve_kv_cache[kv_cache_idx]
|
||||
|
||||
Reference in New Issue
Block a user