[bugfix] compat swift 4.0 (#1680)

examples/pytorch/llm_agent/_common.py

@@ -12,7 +12,7 @@ import numpy as np
 #
 import torch
 from matplotlib.figure import Figure
-from swift import LoRAConfig, Swift
+from swift.tuners import LoRAConfig, Swift
 from tensorboard.backend.event_processing.event_accumulator import \
     EventAccumulator
 from torch import Tensor

modelscope/models/multi_modal/efficient_diffusion_tuning/efficient_stable_diffusion.py

@@ -35,7 +35,7 @@ attention.deprecate = lambda *arg, **kwargs: None
 __tuner_MAP__ = {'lora': LoRATuner, 'control_lora': ControlLoRATuner}
 
 if is_swift_available():
-    from swift import AdapterConfig, LoRAConfig, PromptConfig, Swift
+    from swift.tuners import AdapterConfig, LoRAConfig, PromptConfig, Swift
 
 
 @MODELS.register_module(

modelscope/pipelines/builder.py

@@ -281,7 +281,10 @@ def external_engine_for_llm_checker(model: Union[str, List[str], Model,
 
     llm_framework = kwargs.get('llm_framework', '')
     if llm_framework == 'swift':
-        from swift.llm import get_model_info_meta
+        try:
+            from swift.model import get_model_info_meta
+        except ImportError:
+            from swift.llm import get_model_info_meta
         # check if swift supports
         if os.path.exists(model):
             model_id = get_model_id_from_cache(model)

modelscope/pipelines/nlp/llm_pipeline.py

@@ -217,8 +217,11 @@ class LLMPipeline(Pipeline, PipelineStreamingOutputMixin):
             tokenizer_class) if tokenizer is None else tokenizer
 
     def _init_swift(self, model_id, device) -> None:
-        from swift.llm import prepare_model_template
-        from swift.llm import InferArguments, get_model_info_meta
+        try:
+            from swift.pipelines import prepare_model_template
+            from swift.arguments import InferArguments
+        except ImportError:
+            from swift.llm import prepare_model_template, InferArguments
 
         def format_messages(messages: Dict[str, List[Dict[str, str]]],
                             tokenizer: PreTrainedTokenizer,

modelscope/preprocessors/templates/utils.py

@@ -390,7 +390,10 @@ def load_video_internvl(video_io: BytesIO, bound=None, num_segments=32):
 
 def draw_plot(img_dir: str, bbox: List[int], bbox_type: str, output_file: str):
     from PIL import Image, ImageDraw
-    from swift.llm.template.template import Template
+    try:
+        from swift.template import Template
+    except ImportError:
+        from swift.llm.template.template import Template
     image = Image.open(img_dir)
 
     objects = [{'bbox': bbox, 'bbox_type': bbox_type, 'image': 0}]
@@ -465,7 +468,10 @@ def load_audio_qwen(audio_io: BytesIO, sampling_rate: int):
 
 
 def load_video_qwen2(video_path: str):
-    from swift.llm.template.template import get_env_args
+    try:
+        from swift.utils import get_env_args
+    except ImportError:
+        from swift.llm.template.template import get_env_args
     import torchvision
     from torchvision import io, transforms
     from qwen_vl_utils.vision_process import (round_by_factor, FPS, FRAME_FACTOR, FPS_MIN_FRAMES, FPS_MAX_FRAMES,

modelscope/trainers/hooks/swift/swift_hook.py

@@ -52,7 +52,7 @@ class SwiftCheckpointProcessor(CheckpointProcessor):
                 raise ValueError(
                     'Please install swift by `pip install ms-swift` to use SwiftHook.'
                 )
-            from swift import SwiftModel
+            from swift.tuners import SwiftModel
             if isinstance(model, SwiftModel):
                 _swift_output_dir = output_dir + SwiftCheckpointProcessor.SWIFT_SAVE_SUFFIX
                 model.save_pretrained(

tests/trainers/test_finetune_vision_efficient_tuning_swift.py

@@ -44,7 +44,7 @@ class TestVisionEfficientTuningSwiftTrainer(unittest.TestCase):
         exists('transformers<5.0'),
         'Skip test because transformers version is too high.')
     def test_vision_efficient_tuning_swift_lora_train(self):
-        from swift import LoRAConfig
+        from swift.tuners import LoRAConfig
         model_id = 'damo/cv_vitb16_classification_vision-efficient-tuning-lora'
 
         def cfg_modify_fn(cfg):
@@ -86,7 +86,7 @@ class TestVisionEfficientTuningSwiftTrainer(unittest.TestCase):
         exists('transformers<5.0'),
         'Skip test because transformers version is too high.')
     def test_vision_efficient_tuning_swift_adapter_train(self):
-        from swift import AdapterConfig
+        from swift.tuners import AdapterConfig
         model_id = 'damo/cv_vitb16_classification_vision-efficient-tuning-adapter'
 
         def cfg_modify_fn(cfg):
@@ -126,7 +126,7 @@ class TestVisionEfficientTuningSwiftTrainer(unittest.TestCase):
         exists('transformers<5.0'),
         'Skip test because transformers version is too high.')
     def test_vision_efficient_tuning_swift_prompt_train(self):
-        from swift import PromptConfig
+        from swift.tuners import PromptConfig
         model_id = 'damo/cv_vitb16_classification_vision-efficient-tuning-prompt'
 
         def cfg_modify_fn(cfg):