fix typos

Author: terryyz

README.md

@@ -160,8 +160,8 @@ Following which, you can run the built container as shown in above.
 
 * `task_id` is the identifier string for the task
 * `entry_point` is the name of the function
-* `prompt` is the prompt for BigCodeBench-Complete
-* `instruction` is the prompt for BigCodeBench-Instruct
+* `complete_prompt` is the prompt for BigCodeBench-Complete
+* `instruct_prompt` is the prompt for BigCodeBench-Instruct
 + `canonical_solution` is the ground-truth implementation
 + `test` is the `unittest.TestCase` class
 

bigcodebench/data/bigcodebench.py

@@ -40,12 +40,14 @@ def get_bigcodebench(
     """Get BigCodeBench from BigCode's github repo and return as a list of parsed dicts.
 
     Returns:
-        List[Dict[str, str]]: List of dicts with keys "prompt", "test", "entry_point"
+        List[Dict[str, str]]: List of dicts with keys "complete_prompt", "instruct_prompt", "canonical_solution", "test", "entry_point"
 
     Notes:
         "task_id" is the identifier string for the task.
-        "prompt" is the prompt to be used for the task (function signature with docstrings).
-        "test" is test-cases wrapped in a `check` function.
+        "complete_prompt" is the prompt to be used for BigCodeBench-Complete.
+        "instruct_prompt" is the prompt to be used for BigCodeBench-Instruct.
+        "canonical_solution" is the ground-truth implementation
+        "test" is the `unittest.TestCase` class.
         "entry_point" is the name of the function.
     """
     # Check if open eval file exists in CACHE_DIR

bigcodebench/generate.py

@@ -69,10 +69,7 @@ def codegen(
 
             sidx = n_samples - nsamples
             while sidx < n_samples:
-                if subset == "instruct":
-                    prompt = task["instruct_prompt"]
-                elif subset == "complete":
-                    prompt = task["complete_prompt"]
+                prompt = task[f"{subset}_prompt"]
                 else:
                     raise Exception(f"Invalid subset {subset}")
                 if strip_newlines:

bigcodebench/inspect.py

@@ -30,7 +30,7 @@ def inspection(args):
             os.makedirs(task_path)
         task_id_data = problems[task_id]
         with open(os.path.join(task_path, "ground_truth.py"), "w") as f:
-            f.write(task_id_data["complete_prompt"] + "\n\n" + task_id_data["canonical_solution"])
+            f.write(task_id_data[f"{args.subset}_prompt"] + "\n\n" + task_id_data["canonical_solution"])
 
         # write test
         with open(os.path.join(task_path, "test_case.py"), "w") as f:
@@ -49,6 +49,7 @@ def inspection(args):
 def main():
     parser = argparse.ArgumentParser()
     parser.add_argument("--eval-results", required=True, type=str)
+    parser.add_argument("--subset", required=True, type=str)
     parser.add_argument("--in-place", action="store_true")
     args = parser.parse_args()