Remove ScaledActivation for AWQ (#10057)

Signed-off-by: mgoin <michael@neuralmagic.com>
2024-11-06 09:27:06 -05:00
parent 406d4cc480
commit 399c798608
34 changed files with 19 additions and 124 deletions
--- a/vllm/model_executor/models/bloom.py
+++ b/vllm/model_executor/models/bloom.py
@@ -146,7 +146,7 @@ class BloomMLP(nn.Module):
            4 * hidden_size,
            quant_config=quant_config,
        )
-        self.gelu_impl = get_act_fn("gelu", quant_config, 4 * hidden_size)
+        self.gelu_impl = get_act_fn("gelu")
        self.dense_4h_to_h = RowParallelLinear(
            4 * hidden_size,
            hidden_size,