[Bugfix] Fix Qwen3.5-FP8 nightly fail. Guard fused_add_rms_norm input/weight dtype mismatch in RMSNorm + quant fusion (#44694)
This commit is contained in:
@@ -39,6 +39,7 @@ FP4_DTYPE = torch.uint8
|
|||||||
|
|
||||||
|
|
||||||
_RMS_NORM_OP = torch.ops.vllm_ir.rms_norm.default
|
_RMS_NORM_OP = torch.ops.vllm_ir.rms_norm.default
|
||||||
|
_FUSED_ADD_RMS_NORM_OP = torch.ops.vllm_ir.fused_add_rms_norm.default
|
||||||
|
|
||||||
|
|
||||||
# TODO: extend rmsnorm quant kernels to support mixed input/weight dtypes,
|
# TODO: extend rmsnorm quant kernels to support mixed input/weight dtypes,
|
||||||
@@ -49,8 +50,13 @@ def _rms_input_weight_dtype_match(match: pm.Match) -> bool:
|
|||||||
if node.target == _RMS_NORM_OP:
|
if node.target == _RMS_NORM_OP:
|
||||||
# rms_norm(x, weight, epsilon, variance_size)
|
# rms_norm(x, weight, epsilon, variance_size)
|
||||||
x, weight = node.args[0], node.args[1]
|
x, weight = node.args[0], node.args[1]
|
||||||
if isinstance(x, fx.Node) and isinstance(weight, fx.Node):
|
elif node.target == _FUSED_ADD_RMS_NORM_OP:
|
||||||
return x.meta["val"].dtype == weight.meta["val"].dtype
|
# fused_add_rms_norm(x, residual, weight, epsilon, variance_size)
|
||||||
|
x, weight = node.args[0], node.args[2]
|
||||||
|
else:
|
||||||
|
continue
|
||||||
|
if isinstance(x, fx.Node) and isinstance(weight, fx.Node):
|
||||||
|
return x.meta["val"].dtype == weight.meta["val"].dtype
|
||||||
return True
|
return True
|
||||||
|
|
||||||
|
|
||||||
|
|||||||
Reference in New Issue
Block a user