Upload code/mmsu_inference.py with huggingface_hub

code/mmsu_inference.py

@@ -0,0 +1,80 @@
+import os
+import argparse
+import json
+from tqdm import tqdm
+import shutil
+
+
+def main():
+    # Parse command-line arguments
+    parser = argparse.ArgumentParser(description="Easy Inference for your model.")
+    parser.add_argument('--input_jsonl', type=str, required=True, help="Path to the input JSONL file")
+    parser.add_argument('--output_jsonl', type=str, required=True, help="Path to the output JSONL file")
+    args = parser.parse_args()
+
+    input_file = args.input_jsonl
+    output_file = args.output_jsonl
+
+
+    #Step1: Bulid your model; Implement according to your own model
+    #model = model.cuda()
+    #model.eval()
+
+    #Step2: Single step inference
+    with open(input_file, "r") as fin, open(output_file, "w") as fout:
+        fin = json.load(fin)
+        for item in tqdm(fin):
+            audio_path = item['audio_path']
+            task_name = item['task_name']
+            if os.path.exists(audio_path) == False:
+                print(f"lack wav {wav_fn}")
+                continue
+            
+            
+            #Construct prompt
+            question = item['question']
+            question_prompts = 'Choose the most suitable answer from options A, B, C, and D to respond the question in next line, **you should only choose A or B or C or D.** Do not provide any additional explanations or content.'
+            choice_a = item['choice_a']
+            choice_b = item['choice_b']
+            choice_c = item.get('choice_c', None)
+            choice_d = item.get('choice_d', None)
+            choices = f'A. {choice_a}\nB. {choice_b}\nC. {choice_c}\nD. {choice_d}'
+            instruction = f"{question_prompts}\n\nQuestion: {question}\n\n{choices}"
+            
+
+            #Step 3: Run model inference
+            #output = model.infer(
+            #    Prompts=instruction,
+            #    Audio_path=audio_path,
+            #    ...
+            #)
+            
+            output = "Model response here"  # Placeholder response
+
+            #Step 4: save result
+            json_string = json.dumps(
+                {
+                    "id": item["id"]
+                    "audio_path": item["audio_path"],
+                    "question": question,
+                    "choice_a": choice_a,
+                    "choice_b": choice_b,
+                    "choice_c": choice_c,
+                    "choice_d": choice_d,
+                    "answer_gt": item["answer_gt"],
+                    "response": output,
+                    "task_name": task_name,
+                    "category": item["category"],
+                    "sub-category": item["category"],
+                    "sub-sub-category": item["sub-sub-category"],
+                    "linguistics_sub_discipline": item["linguistics_sub_discipline"],
+                },
+                ensure_ascii=False
+            )
+            fout.write(json_string + "\n")
+            
+
+if __name__ == "__main__":
+    # bash:
+    # python mmsu_inference.py --input_jsonl /path/to/input.jsonl --output_jsonl /path/to/output.jsonl
+    main()