Upload folder using huggingface_hub

server/app.py

@@ -58,7 +58,10 @@ def get_llm_solution(problem_desc: str):
             text = text.split("```rust")[1].split("```")[0]
         elif "```" in text:
             text = text.split("```")[1].split("```")[0]
-        return text.strip()
+        text = text.strip()
+        if not text:
+            return "// LLM Error: empty response (no code returned)."
+        return text
     except Exception as e:
         return f"// LLM Error: {e}"
 
@@ -72,7 +75,7 @@ def evaluate_single(problem_id, code=None):
         solution_code = code if code else get_llm_solution(problem["description"])
 
         # 2. Guard: If LLM failed, do not evaluate
-        if solution_code.startswith("// LLM Error"):
+        if not solution_code.strip() or solution_code.startswith("// LLM Error"):
             return solution_code, {"error": "LLM failed to generate a solution. Check your HF_TOKEN."}
 
         # 3. Evaluate properly

server/rust_coder_environment.py

@@ -118,16 +118,21 @@ class RustCoderEnvironment(Environment):
         code = action.code
 
         if not code.strip():
-            done = self.current_problem_idx >= len(self.problems) - 1
-            if not done:
-                self.current_problem_idx += 1
+            # Invalid/empty submission: do not advance the problem index.
+            done = False
             return RustCoderObservation(
                 problem_description=problem["description"],
                 starter_code=problem.get("starter_code", ""),
                 compilation_success=False,
                 compilation_output="Error: no code submitted.",
                 test_results=[],
-                reward_breakdown={"compilation": 0.0, "correctness": 0.0, "coverage": 0.0, "elegance": 0.0, "efficiency": 0.0},
+                reward_breakdown={
+                    "compilation": 0.0,
+                    "correctness": 0.0,
+                    "coverage": 0.0,
+                    "elegance": 0.0,
+                    "efficiency": 0.0,
+                },
                 done=done,
                 reward=0.0,
             )
@@ -165,11 +170,11 @@ class RustCoderEnvironment(Environment):
 
         # ── Total reward ──────────────────────────────────────────────
         reward_breakdown = {
-            "Compilation":  round(r_compilation,  4),
-            "Correctness":  round(r_correctness,  4),
-            "Coverage":     round(r_coverage,     4),
-            "Elegance":     round(r_elegance,     4),
-            "Efficiency":   round(r_efficiency,   4),
+            "compilation":  round(r_compilation,  4),
+            "correctness":  round(r_correctness,  4),
+            "coverage":     round(r_coverage,     4),
+            "elegance":     round(r_elegance,     4),
+            "efficiency":   round(r_efficiency,   4),
         }
         # Calculate weighted total reward
         total_reward = round(