Upload infer_qwen2_vl.py with huggingface_hub

infer_qwen2_vl.py

@@ -42,7 +42,7 @@ file_names = os.listdir(folder)
 
 num_image = len(file_names)
 
-begin, end, batch_size= 0, 10, 2
+begin, end, batch_size= 0, num_image, 4
 print(f"beigin : {begin}, end : {end}, batch_size : {batch_size}")
 messages = [
     {
@@ -73,9 +73,9 @@ for batch_idx in tqdm(range(begin, end, batch_size)):
     image_inputs_batch, video_inputs_batch,text_batch = [], [], []
     for idx,i in enumerate(batch):
         #img = batch[i]
-        print('gain image successfully !')
+        #print('gain image successfully !')
         img_path = '/inspire/hdd/ws-ba572160-47f8-4ca1-984e-d6bcdeb95dbb/a100-maybe/albus/ICCV_2025/qvq/dataset/' + i
-        print(img_path)
+        #print(img_path)
         messages[1]["content"][0]["image"] = '/inspire/hdd/ws-ba572160-47f8-4ca1-984e-d6bcdeb95dbb/a100-maybe/albus/ICCV_2025/qvq/dataset/' + i
         text = processor.apply_chat_template(
             messages, tokenize=False, add_generation_prompt=True
@@ -117,7 +117,7 @@ for batch_idx in tqdm(range(begin, end, batch_size)):
         generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
     )
     ans.append(output_text)
-    save_path = "output_1.json"
+    save_path = "output_final.json"
     counter = counter + 1
     if counter % 1 == 0:
         print(f"Saving data at iteration {idx + 1}")