fix(eval): SWE-Bench instance with upper-case instance id (#4649)

evaluation/swe_bench/eval_infer.py

@@ -239,7 +239,7 @@ def process_instance(
                         # Create a directory structure that matches the expected format
                         # NOTE: this is a hack to make the eval report format consistent
                         # with the original SWE-Bench eval script
-                        log_dir = os.path.join(temp_dir, 'logs', instance_id)
+                        log_dir = os.path.join(temp_dir, 'logs', instance_id.lower())
                         os.makedirs(log_dir, exist_ok=True)
                         test_output_path = os.path.join(log_dir, 'test_output.txt')
                         with open(test_output_path, 'w') as f:

evaluation/swe_bench/run_infer.py

@@ -101,7 +101,7 @@ def get_instance_docker_image(instance_id: str) -> str:
     image_name = image_name.replace(
         '__', '_s_'
     )  # to comply with docker image naming convention
-    return DOCKER_IMAGE_PREFIX.rstrip('/') + '/' + image_name
+    return (DOCKER_IMAGE_PREFIX.rstrip('/') + '/' + image_name).lower()
 
 
 def get_config(