support moe offline quant (#5142)

This commit is contained in:
xiaoxiaohehe001
2025-11-24 18:59:18 +08:00
committed by GitHub
parent 5ff93d4998
commit e150a418d4
5 changed files with 12 additions and 3 deletions
+1 -1
View File
@@ -584,7 +584,7 @@ class FusedMoE(nn.Layer):
"""
load_state_dict function.
"""
if self.is_quantized:
if self.is_quantized or self.fd_config.model_config.is_moe_quantized:
if getattr(self.fd_config.quant_config, "is_permuted", True):
self.quant_method.process_prequanted_weights(self, state_dict, is_rearrange)
else: