Upload main.py with huggingface_hub

main.py

@@ -43,7 +43,8 @@ app.add_middleware(
 VALID_TASKS = [
     "easy_dedup_rename",
     "medium_missing_dtype",
-    "hard_full_pipeline"
+    "hard_full_pipeline",
+    "expert_sales_pipeline"
 ]
 
 envs: Dict[str, DataCleaningEnv] = {
@@ -135,6 +136,21 @@ def list_tasks():
                     "remove_duplicates", "fill_missing", "fix_dtype",
                     "remove_outliers", "validate_schema", "finish"
                 ]
+            },
+            {
+                "task_id":   "expert_sales_pipeline",
+                "difficulty": "expert",
+                "description": (
+                    "Expert level: Full sales data cleaning pipeline "
+                    "requiring correct order of operations including "
+                    "case standardization, outlier removal, and schema validation."
+                ),
+                "max_steps": 25,
+                "operations": [
+                    "remove_duplicates", "fill_missing", "fix_dtype",
+                    "remove_outliers", "rename_columns",
+                    "validate_schema", "finish"
+                ]
             }
         ]
     }
@@ -231,3 +247,75 @@ def validate():
         "openenv_valid": all_passed,
         "tasks": results
     }
+
+
+# In memory leaderboard
+leaderboard_data = []
+
+@app.post("/leaderboard/submit")
+def submit_score(entry: Dict[str, Any]):
+    """Submit a score to the leaderboard."""
+    required = ["model_name", "task_id", "score"]
+    for field in required:
+        if field not in entry:
+            raise HTTPException(
+                status_code=400,
+                detail=f"Missing field: {field}"
+            )
+    if not 0.0 <= float(entry["score"]) <= 1.0:
+        raise HTTPException(
+            status_code=400,
+            detail="Score must be between 0.0 and 1.0"
+        )
+    leaderboard_data.append({
+        "model_name": entry["model_name"],
+        "task_id":    entry["task_id"],
+        "score":      round(float(entry["score"]), 4),
+        "steps":      entry.get("steps", 0),
+        "timestamp":  __import__("datetime").datetime.utcnow().isoformat()
+    })
+    return {"status": "submitted", "entry": leaderboard_data[-1]}
+
+
+@app.get("/leaderboard")
+def get_leaderboard():
+    """Get current leaderboard rankings."""
+    if not leaderboard_data:
+        # Return baseline scores
+        return {
+            "leaderboard": [
+                {
+                    "rank": 1,
+                    "model_name": "gpt-4o-mini (baseline)",
+                    "easy_score":   1.0000,
+                    "medium_score": 0.6643,
+                    "hard_score":   0.8386,
+                    "avg_score":    0.8343
+                }
+            ],
+            "total_submissions": 1
+        }
+
+    # Group by model
+    from collections import defaultdict
+    model_scores = defaultdict(dict)
+    for entry in leaderboard_data:
+        model_scores[entry["model_name"]][entry["task_id"]] = entry["score"]
+
+    ranked = []
+    for model, scores in model_scores.items():
+        avg = sum(scores.values()) / len(scores) if scores else 0
+        ranked.append({
+            "model_name":   model,
+            "scores":       scores,
+            "avg_score":    round(avg, 4)
+        })
+
+    ranked.sort(key=lambda x: x["avg_score"], reverse=True)
+    for i, r in enumerate(ranked):
+        r["rank"] = i + 1
+
+    return {
+        "leaderboard":       ranked,
+        "total_submissions": len(leaderboard_data)
+    }