fix: add extra name in save ppath

Author: terryyz

bigcodebench/data/bigcodebench.py

@@ -27,13 +27,13 @@ def _ready_bigcodebench_path(subset="full", version="default") -> str:
 
     extra = "-" + subset if subset else ""
 
-    try:
-        dataset = load_dataset(BIGCODEBENCH_HF+extra, split=BIGCODEBENCH_VERSION)
-        make_cache(url, dataset, path)
-    except:
-        if os.path.exists(path):
-            os.remove(path)
-        make_cache(url, None, path, gh=True)
+    # try:
+    dataset = load_dataset(BIGCODEBENCH_HF+extra, split=BIGCODEBENCH_VERSION)
+    make_cache(url, dataset, path)
+    # except:
+    #     if os.path.exists(path):
+    #         os.remove(path)
+    #     make_cache(url, None, path, gh=True)
 
     return path
 

bigcodebench/data/utils.py

@@ -20,9 +20,8 @@ def get_dataset_metadata(version: str, subset: str="full"):
 
 def make_cache(gzip_url, hf_data, cache_path, gh=False):
     # Check if open eval file exists in CACHE_DIR
-    if not os.path.exists(cache_path):
-        
-        if gh:
+    if gh:
+        if not os.path.exists(cache_path):
             # Install BigCodeBench dataset and parse as jsonl
             print(f"Downloading dataset from {gzip_url}")
             with tempdir.TempDir() as tmpdir:
@@ -39,8 +38,8 @@ def make_cache(gzip_url, hf_data, cache_path, gh=False):
             # Write the original open eval file to CACHE_DIR
             with open(cache_path, "w") as f:
                 f.write(data)
-        else:
-            hf_data.to_json(cache_path)
+    else:
+        hf_data.to_json(cache_path)
 
 
 def write_jsonl(

bigcodebench/generate.py

@@ -148,7 +148,7 @@ def main():
 
     extra = "-" + args.subset if args.subset else ""
     if not args.save_path:
-        save_path = args.model.replace("/", "--") + f"--bigcodebench-{args.split}--{args.backend}-{args.temperature}-{args.n_samples}.jsonl"
+        save_path = args.model.replace("/", "--") + f"--bigcodebench{extra}-{args.split}--{args.backend}-{args.temperature}-{args.n_samples}.jsonl"
     else:
         save_path = args.save_path