Falcon: Enable fused attention

f03f72de · Casper Hansen · ac770875 · f03f72de
Commit f03f72de authored Sep 12, 2023 by Casper Hansen
Show whitespace changes
Inline Side-by-side

Showing with 1 addition and 1 deletion

awq/models/falcon.py awq/models/falcon.py +1 -1

No files found.
--- a/awq/models/falcon.py
+++ b/awq/models/falcon.py
@@ -7,7 +7,7 @@ class FalconAWQForCausalLM(BaseAWQForCausalLM):
    @staticmethod
    def fuse_layers(model: FalconForCausalLM, quant_config:dict):
        fuser = FalconFuser(model)
-        # fuser.fuse_transformer()
+        fuser.fuse_transformer()

    @staticmethod
    def get_model_layers(model: FalconForCausalLM):