chore: add extra empty cache

This commit is contained in:
NanoCode012
2026-02-25 18:17:11 +07:00
parent 731d5dd193
commit c58eaaae51

View File

@@ -197,6 +197,7 @@ class ModelLoader:
if get_moe_quantized_count() > 0:
self.model._moe_experts_quantized = True
patch_peft_target_parameters_matching()
torch.cuda.empty_cache()
PLUGIN_MANAGER.post_model_build(self.cfg, self.model)