Upload folder using huggingface_hub

inference.py

@@ -199,17 +199,39 @@ def run_task(task_id: str) -> float:
 
 def main():
     specific_task = TASK_ID
+    results_dir = "outputs"
+    os.makedirs(results_dir, exist_ok=True)
+    results_path = os.path.join(results_dir, "baseline_results.json")
+
+    final_data = {
+        "model": MODEL_NAME,
+        "timestamp": time.strftime("%Y-%m-%d %H:%M:%S"),
+        "tasks": {}
+    }
+
     if specific_task and specific_task in ALL_TASKS:
-        run_task(specific_task)
+        score = run_task(specific_task)
+        final_data["tasks"][specific_task] = {"score": score}
     else:
         # Run all tasks so the validator sees graders for every task
         all_scores = []
         for t_id in ALL_TASKS:
             score = run_task(t_id)
             all_scores.append(score)
+            final_data["tasks"][t_id] = {"score": score}
+        
         avg = sum(all_scores) / len(all_scores)
+        final_data["avg_score"] = avg
         print(f"[SUMMARY] tasks={len(ALL_TASKS)} avg_score={avg:.4f}", flush=True)
 
+    # Save to JSON for local tracking
+    try:
+        with open(results_path, "w") as f:
+            json.dump(final_data, f, indent=2)
+        print(f"[DEBUG] Results saved to {results_path}", flush=True)
+    except Exception as e:
+        print(f"[DEBUG] Could not save progress to JSON: {e}", flush=True)
+
 
 if __name__ == "__main__":
     main()