[Bugfix] Fix nemotron accuracy drop introduced by #41184 (#45037)

Signed-off-by: Bill Nell <bnell@redhat.com>
This commit is contained in:
bnellnm
2026-06-10 13:16:25 -04:00
committed by GitHub
parent dc66e01a70
commit 29026682cb
@@ -349,7 +349,9 @@ def FusedMoE(
topk_group=topk_group,
custom_routing_function=custom_routing_function,
scoring_func=scoring_func,
routed_scaling_factor=routed_scaling_factor,
routed_scaling_factor=routed_scaling_factor
if not apply_routed_scale_to_output
else 1.0,
swiglu_limit=swiglu_limit,
# TODO get from router? needs to be truncated?
e_score_correction_bias=e_score_correction_bias,