Skip to content

GitLab

Unverified Commit 7802586c authored by Rain Jiang's avatar Rain Jiang Committed by GitHub
Browse files

fix the fp8 topk_config.correction_bias is none bug (#10040)

parent bc5fc332
Hide whitespace changes
Inline Side-by-side
Showing with 6 additions and 4 deletions
python/sglang/srt/layers/quantization/fp8.py
View file @ 7802586c
...@@ -1132,10 +1132,12 @@ class Fp8MoEMethod(FusedMoEMethodBase): ...@@ -1132,10 +1132,12 @@ class Fp8MoEMethod(FusedMoEMethodBase):
and topk_config.topk_group is not None and topk_config.topk_group is not None
), "Current trtllm_fp8_block_scale_moe kernel does not support these two arguments as None" ), "Current trtllm_fp8_block_scale_moe kernel does not support these two arguments as None"
if topk_config.correction_bias is None: correction_bias = (
correction_bias = topk_config.correction_bias.to(x.dtype) None
else: if topk_config.correction_bias is None
correction_bias = None else topk_config.correction_bias.to(x.dtype)
)
return trtllm_fp8_block_scale_moe( return trtllm_fp8_block_scale_moe(
routing_logits=router_logits.to(torch.float32), routing_logits=router_logits.to(torch.float32),
routing_bias=correction_bias, routing_bias=correction_bias,
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment