A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 15, 01:00 UTC
[Bugfix] Fix Machete zero point issue for GPTQ models on SM90 (#21066)
Signed-off-by: mgoin <mgoin64@gmail.com>
Michael Goin committed
Jul 16, 2025 at 22:54 UTC
28a6d5423db63ba9c4df13608f6151a484bdb7c9
1 file changed
+5
vllm/model_executor/layers/quantization/kernels/mixed_precision/machete.py
+5
@@ -126,6 +126,11 @@ class MacheteLinearKernel(MPLinearKernel):
126
if c.has_g_idx:
127
x_2d = self.act_perm(x_2d)
128
129
+ if c.zero_points:
130
+ assert w_zp is not None
131
+ else:
132
+ w_zp = None
133
+
134
output = ops.machete_mm(a=x_2d,
135
b_q=w_q,
136
b_type=c.weight_type,