update files

processing_logics.py

@@ -57,7 +57,6 @@ def process_video_with_decord(video_file, data_args):
         video = torch.zeros((10, 720, 720, 3)).numpy()
         print(f"load {video_file} error, use empty tensor instead.")
     frame_time = ",".join([f"{i:.2f}s" for i in frame_time])
-    print("load video : \t", video.shape)
     num_frames_to_sample = num_frames = len(frame_idx)
     # https://github.com/dmlc/decord/issues/208
     vr.seek(0)
@@ -231,17 +230,9 @@ def preprocess_qwen(sources, tokenizer: PreTrainedTokenizer,enable_thinking: boo
         conv = {"role" : role, "content" : content}
         messages.append(conv)
 
-    
-    print(f"enable_thinking: {enable_thinking}")
     input_id = tokenizer.apply_chat_template(messages, add_generation_prompt=True, enable_thinking=enable_thinking)
     
 
-    print(f"---------------------------------")
-    print(f"****COMPETE_CONV****\n{messages}")
-    print(f"****TOKENIZE_RESULTS****\n{tokenizer.apply_chat_template(messages,tokenize=False,add_generation_prompt=True)}")
-    print(f"****INPUT_IDS****\n{input_id}")
-    print(f"---------------------------------")
-
     for idx, encode_id in enumerate(input_id):
         if encode_id == image_token_index:
             input_id[idx] = IMAGE_TOKEN_INDEX
@@ -325,7 +316,6 @@ class LogicsProcessor(ProcessorMixin):
         text: Union[TextInput, PreTokenizedInput, list[TextInput], list[PreTokenizedInput]] = None,
         **kwargs
     ) -> BatchFeature:
-        print(text)
         device='cuda'
         if not isinstance(images, list):
             images = [images]
@@ -364,14 +354,12 @@ class LogicsProcessor(ProcessorMixin):
             image_inputs["image_sizes"] = [video_frames[0].size]
             image_inputs["modalities"] = ["video"]
 
-            print(image_inputs)
             
 
         # ---------------- text ----------------
         text_ids = self.text_tokenizer(question=text, img_path=images, tokenizer=self.tokenizer, enable_thinking=self.config.enable_thinking)
         text_inputs = {}
         text_inputs["text_inputs"] = text_ids
-        print(text_inputs)
 
         if images is not None:
             return BatchFeature(data={**text_inputs, **image_inputs})