https://github.com/FasterDecoding/SnapKV/blob/ea655b18061313e088879bd2b4a3e3c0c2dc2e21/snapkv_utils.py#L50 In `update_kv` function, instead of using the function's arguments `attention_mask`, this variable is overridden.
SnapKV/snapkv_utils.py
Line 50 in ea655b1
In
update_kvfunction, instead of using the function's argumentsattention_mask, this variable is overridden.