[BUGFIX][Quant]Fix test_kv_scale_reload failed (#50405)
Signed-off-by: Lai, Yejing <yejing.lai@intel.com> Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
Yejing Lai committed
Aug 5, 2026 at 10:50 UTC
2cb3ff88cc6b0cd1e4f2eb96530b1664d4accb6b
1 file changed
+3
-2
vllm/model_executor/layers/quantization/compressed_tensors/schemes/compressed_tensors_w8a16_fp8.py
+3
-2
@@ -137,8 +137,9 @@ class CompressedTensorsW8A16Fp8(CompressedTensorsScheme):
137
"weight_scale",
138
convert_to_channelwise(layer.weight_scale, layer.logical_widths),
139
)
140
- self.strategy = QuantizationStrategy.CHANNEL
141
- self.weight_quant_key = STRATEGY_TO_WEIGHT_QUANT_KEY[self.strategy]
140
+ self.weight_quant_key = STRATEGY_TO_WEIGHT_QUANT_KEY[
141
+ QuantizationStrategy.CHANNEL
142
+ ]
143
self.linear_kernel.config.weight_quant_key = self.weight_quant_key
144
145
# Canonicalize to (K, N) for the kernel.