pip 0.3.4 (#697)

* Update project version to 0.3.4 and enhance setup for license file inclusion in setup.py

* lint

Author: pufanyi

lmms_eval/api/task.py

@@ -952,7 +952,7 @@ def _download_from_youtube(path):
                     )
                     zip_files = glob(os.path.join(cache_path, "**/*.zip"), recursive=True)
                     tar_files = glob(os.path.join(cache_path, "**/*.tar*"), recursive=True)
-                    
+
                     def unzip_video_data(zip_file):
                         import os
                         import zipfile
@@ -995,7 +995,7 @@ def concat_tar_parts(tar_parts, output_tar):
                         # Group tar parts together
                         for tar_file in tar_files:
                             base_name = tar_file.split(".tar")[0]
-                            base_name = re.sub(r'_\d+$', '', base_name)
+                            base_name = re.sub(r"_\d+$", "", base_name)
                             if base_name not in tar_parts_dict:
                                 tar_parts_dict[base_name] = []
                             tar_parts_dict[base_name].append(tar_file)

lmms_eval/tasks/videoevalpro/utils.py

@@ -3,16 +3,16 @@
 import os
 import re
 import sys
+import time
 from collections import defaultdict
 from pathlib import Path
 from typing import Dict, List, Optional, Union
 
 import cv2
 import numpy as np
+import openai
 import yaml
 from loguru import logger as eval_logger
-import openai
-import time
 
 from lmms_eval.tasks._task_utils.file_utils import generate_submission_file
 
@@ -71,19 +71,12 @@ def __call__(self, question: str, target: str, predicted_answer: str) -> bool:
 Grade the predicted answer ofthe question as one of: A: CORRECT B: INCORRECT C: NOT_ATTEMPTED Just return the letter "A", "B", or "C", with no text around it.
 """.strip()
 
-        response = self.client.chat.completions.create(
-            model=self.model_name,
-            messages=[
-                {"role": "user", "content": prompt}
-            ],
-            temperature=0,
-            max_tokens=5
-        )
+        response = self.client.chat.completions.create(model=self.model_name, messages=[{"role": "user", "content": prompt}], temperature=0, max_tokens=5)
 
         answer = response.choices[0].message.content.strip()
         # print(f"---------------------->>>>gpt answer: {answer}")
         return answer[0].upper() == "A"
-    
+
 
 def safe_judge_with_retry(model, question, gt, pred, max_retries=10, delay=2):
     for attempt in range(max_retries):
@@ -95,9 +88,11 @@ def safe_judge_with_retry(model, question, gt, pred, max_retries=10, delay=2):
     print(f"All {max_retries} attempts failed.")
     return {"correct": False, "reason": "Failed after multiple retries"}
 
+
 def safe_strip(x):
     return x.strip() if isinstance(x, str) else ""
 
+
 def videoevalpro_process_results(doc, results):
     """
     Args:
@@ -141,14 +136,14 @@ def videoevalpro_aggregate_results(results):
         # overall score
         category2score["overall"]["answered"] += 1
         category2score["overall"]["correct"] += result["judge_result"] == True
-    
+
     final_scores = {}
     for task_type, score in category2score.items():
         if score["answered"] > 0:
             final_scores[task_type] = score["correct"] / score["answered"]
         else:
             final_scores[task_type] = 0.0
-    
+
     eval_logger.info(f"Final scores: {final_scores}")
-  
+
     return final_scores

pyproject.toml

@@ -7,19 +7,19 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "lmms_eval"
-version = "0.3.0"
+version = "0.3.4"
 authors = [
     { name = "LMMMs-Lab Evaluation Team", email = "[email protected]" },
 ]
 description = "A framework for evaluating large multi-modality language models"
 readme = "README.md"
+license = { file = "LICENSE" }
 classifiers = [
     "Programming Language :: Python :: 3",
     "License :: OSI Approved :: MIT License",
     "Operating System :: OS Independent",
 ]
 requires-python = ">=3.12"
-license = { text = "MIT" }
 dependencies = [
     "accelerate>=0.29.1",
     "black>=24.1.0",

setup.py

@@ -1,4 +1,8 @@
 import setuptools
+from setuptools import setup
 
 # This is to make sure that the package supports editable installs
-setuptools.setup()
+if __name__ == "__main__":
+    setuptools.setup(
+        license_files=["LICENSE"],
+    )