From ca0d92227e3a5e5880dde67da9d96c6d06454328 Mon Sep 17 00:00:00 2001 From: Michael Goin Date: Fri, 25 Oct 2024 15:40:33 -0400 Subject: [PATCH] [Bugfix] Fix compressed_tensors_moe bad config.strategy (#9677) --- .../quantization/compressed_tensors/compressed_tensors_moe.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors_moe.py b/vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors_moe.py index 733eece4b5fa..c21aaa40ff2c 100644 --- a/vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors_moe.py +++ b/vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors_moe.py @@ -245,7 +245,7 @@ class CompressedTensorsWNA16MoEMethod(CompressedTensorsMoEMethod): config = self.quant_config.target_scheme_map["Linear"].get("weights") self.num_bits = config.num_bits self.packed_factor = 32 // config.num_bits - self.strategy = config.strategy.value + self.strategy = config.strategy self.group_size = config.group_size assert config.symmetric, ( "Only symmetric quantization is supported for MoE")