fix scedit bug (#290)

Author: tastelikefeet

swift/llm/sft.py

@@ -234,7 +234,9 @@ def llm_sft(args: SftArguments) -> Dict[str, Union[str, Any]]:
     class TrainerAdapterCallback(TrainerCallback):
 
         def on_train_begin(*args, **kwargs):
-            model.set_active_adapters(model.adapters.keys(), offload='meta')
+            if hasattr(model, 'set_active_adapters'):
+                model.set_active_adapters(
+                    model.adapters.keys(), offload='meta')
 
     trainer = Seq2SeqTrainer(
         model=model,

swift/tuners/base.py

@@ -48,10 +48,12 @@ def __init__(self,
             extra_state_keys.extend(model.extra_state_keys)
             model = model.base_model
 
+        new_adapters = []
         if isinstance(config, SwiftConfig):
             if DEFAULT_ADAPTER not in self.adapters:
                 self.adapters[DEFAULT_ADAPTER] = self._prepare_model(
                     model, config, DEFAULT_ADAPTER)
+                new_adapters.append(DEFAULT_ADAPTER)
             else:
                 logger.warn(
                     f'Adater {DEFAULT_ADAPTER} has been patched, skip.')
@@ -61,6 +63,7 @@ def __init__(self,
                 if adapter_name not in self.adapters:
                     self.adapters[adapter_name] = self._prepare_model(
                         model, _config, adapter_name)
+                    new_adapters.append(adapter_name)
                 else:
                     logger.warn(
                         f'Adater {adapter_name} has been patched, skip.')
@@ -76,14 +79,15 @@ def forward(self, *args, **kwargs):
             signature(self.base_model.forward).parameters.values())
         forward.__signature__ = Signature(_parameters)
         self.forward = MethodType(forward, self)
-        for adapter_name in self.adapters:
+        for adapter_name in new_adapters:
             self.activate_adapter(adapter_name)
 
         if inference_mode:
             self.eval()
         else:
-            for output in self.adapters.values():
-                output.mark_trainable_callback(model)
+            for key, output in self.adapters.items():
+                if key in new_adapters:
+                    output.mark_trainable_callback(model)
             if self.extra_state_keys:
                 for n, p in model.named_parameters():
                     if any(

swift/tuners/scetuning/scetuning.py

@@ -205,6 +205,7 @@ def _forward_decoder_mode(self, *args, **kwargs):
             tuner_op = SCETunerModule(
                 name=config.tuner_op,
                 adapter_name=adapter_name,
+                module_key=str(tuner_id),
                 dim=dims[tuner_id],
                 tuner_length=int(dims[tuner_id] * config.down_ratio))
             setattr(t_module, f'scetuner_{adapter_name}', tuner_op)
@@ -244,6 +245,7 @@ class SCETunerModule(nn.Module, ActivationMixin):
     def __init__(self,
                  name,
                  adapter_name,
+                 module_key,
                  dim,
                  tuner_length,
                  tuner_type=None,
@@ -252,7 +254,7 @@ def __init__(self,
                  zero_init_last=True,
                  use_bias=True):
         super(SCETunerModule, self).__init__()
-        super(nn.Module, self).__init__('')
+        super(nn.Module, self).__init__(module_key)
         self.name = name
         self.adapter_name = adapter_name
         self.dim = dim
@@ -271,6 +273,7 @@ def forward(self, x, x_shortcut=None, use_shortcut=True, **kwargs):
         if not self.is_activated(self.adapter_name):
             return x
         if self.name == 'SCEAdapter':
+            self.tuner_op.to(x.device)
             out = self.tuner_op(x)
         else:
             raise Exception(f'Error tuner op {self.name}')