fix

Author: terryyz

bigcodebench/evaluate.py

@@ -231,7 +231,7 @@ def evaluate(
         problems = get_bigcodebench(subset=subset)
 
         # Add selective evaluation logic
-        if selective_evaluate:
+        if selected_ids:
             problems = {k: v for k, v in problems.items() if k in selected_ids}
             if not problems:
                 raise ValueError(f"None of the provided task IDs {selected_ids} were found in the dataset")

sandbox-templates/e2b.Dockerfile

@@ -28,7 +28,7 @@ RUN adduser --disabled-password --gecos "" bigcodebenchuser
 RUN rm -rf /bigcodebench
 
 # Acquire benchmark code to local
-RUN echo 2
+RUN echo 3
 ADD "https://api.github.com/repos/bigcode-project/bigcodebench/commits?per_page=1" latest_commit
 RUN git clone https://github.com/bigcode-project/bigcodebench.git /bigcodebench