diff --git a/train.py b/train.py
index 13421d9..7b4e045 100644
--- a/train.py
+++ b/train.py
@@ -176,6 +176,7 @@ def train(config_path):
             "activation_checkpointing": True,
             "limit_all_gathers": True,
             "sync_module_states": True,
+            "use_orig_params": True,  # Critical for LoRA/PEFT
             "mixed_precision": {"param_dtype": torch.bfloat16, "reduce_dtype": torch.float32, "buffer_dtype": torch.float32},
             "auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
             "transformer_layer_cls_to_wrap": "Qwen3_5MoeDecoderLayer",