[Bugfix] Skip linear bias in layerwise reload to avoid corruption (#49586)

Signed-off-by: li-jinpeng <3332126450@qq.com>
Signed-off-by: xymli <xymli@tencent.com>
Co-authored-by: xymli <xymli@tencent.com>
This commit is contained in:
li-jinpeng
2026-07-24 16:53:48 +08:00
committed by GitHub
co-authored by xymli
parent 589a5b884b
commit 163ecba377
@@ -29,6 +29,9 @@ SKIP_TENSORS: set[str] = {
"expert_physical_to_global",
"expert_local_to_global",
"e_score_correction_bias",
# Built after create_weights(), so it is not tracked by the layerwise-reload
# trigger and would be re-materialized into uninitialized memory. Skip it.
"bias",
}