Upload folder using huggingface_hub

inference.py

@@ -29,11 +29,17 @@ from code_review.client import CodeReviewEnv
 API_BASE_URL = "https://router.huggingface.co/v1"
 API_KEY = os.getenv("HF_TOKEN")
 MODEL_NAME = os.getenv("MODEL_NAME")
+TASK_NAME = "code_review"
+BENCHMARK = "code_review_benchmark"
 MAX_STEPS = 3
 TEMPERATURE = 0.2
-MAX_TOKENS = 512
+MAX_TOKENS = 150
+NUM_EPISODES = 6
+_MAX_REWARD_PER_STEP = MAX_TOKENS * 0.1
+MAX_TOTAL_REWARD = NUM_EPISODES * MAX_STEPS * _MAX_REWARD_PER_STEP
+SUCCESS_SCORE_THRESHOLD = 0.1  # normalized score in [0, 1]
+
 
-DEBUG = True
 ACTION_PREFIX_RE = re.compile(
     r"^(action|next action)\s*[:\-]\s*",
     re.IGNORECASE,
@@ -195,12 +201,12 @@ def parse_action(text: str) -> Dict[str, Any]:
 
 async def run_episode(client, env):
     result = await env.reset()
-
     obs = result.observation
-
     final_score = 0.0
 
     for step in range(1, MAX_STEPS + 1):
+        if result.done:
+            break
 
         prompt = build_prompt(step, MAX_STEPS, obs)
 
@@ -210,15 +216,12 @@ async def run_episode(client, env):
         ]
 
         completion = safe_completion(client, messages)  # still sync
-        # print(completion)
         if completion is None:
             action = fallback_action()
         else:
             response_text = completion.choices[0].message.content or ""
             action_dict = parse_action(response_text)
 
-            # print(response_text)
-
             action = CodeReviewAction(
                 action_type=action_dict.get("action_type"),
                 comment=action_dict.get("comment"),
@@ -227,20 +230,14 @@ async def run_episode(client, env):
             )
 
         result = await env.step(action)
-        # print("Result === " , result)
 
         obs = result.observation
         reward = result.reward
         done = result.done
 
+        log_step(step=step, action=response_text, reward=reward.score, done=done, error=None)
         final_score = max(final_score, reward.score if reward else 0.0)
 
-        print(f"Step {step} | Action: {action} | Reward: {reward}")
-
-        if done:
-            print(f"Done in {step} steps")
-            break
-
     return final_score
 
 
@@ -248,23 +245,22 @@ async def main():
     client = OpenAI(base_url=API_BASE_URL, api_key=API_KEY)
 
     scores = []
-    # log_start(task=TASK_NAME, env=BENCHMARK, model=MODEL_NAME)
+    log_start(task=TASK_NAME, env=BENCHMARK, model=MODEL_NAME)
 
     async with CodeReviewEnv(base_url="http://localhost:8000") as env:
-        NUM_EPISODES = 6
-
         for i in range(NUM_EPISODES):
-            print(f"\n=== Episode {i+1} ===")
+            print(f"\n===== Episode {i+1} =====", flush=True)
             env.task_index = i
 
             score = await run_episode(client, env)
             scores.append(score)
 
-            print(f"Scores so far: {scores}")
-            # return 0
+            # print(f"[INFO] Scores so far: {scores}", flush=True)
 
-    print("\nFinished all episodes")
-    print(f"Final Scores: {scores}")
+    total_score = sum(scores) / MAX_TOTAL_REWARD if MAX_TOTAL_REWARD > 0 else 0.0
+    final_score = min(max(score, 0.0), 1.0)  # clamp to [0, 1]
+    success = final_score >= SUCCESS_SCORE_THRESHOLD
+    log_end(success=success, steps=NUM_EPISODES*MAX_STEPS, score=final_score, rewards=scores)
 
 
 if __name__ == "__main__":

openenv_code_review.egg-info/PKG-INFO

@@ -0,0 +1,9 @@
+Metadata-Version: 2.4
+Name: openenv-code_review
+Version: 0.1.0
+Summary: Code Review environment for OpenEnv
+Requires-Python: >=3.10
+Requires-Dist: openenv-core[core]>=0.2.1
+Provides-Extra: dev
+Requires-Dist: pytest>=8.0.0; extra == "dev"
+Requires-Dist: pytest-cov>=4.0.0; extra == "dev"

openenv_code_review.egg-info/SOURCES.txt

@@ -0,0 +1,15 @@
+README.md
+pyproject.toml
+./__init__.py
+./client.py
+./inference.py
+./models.py
+openenv_code_review.egg-info/PKG-INFO
+openenv_code_review.egg-info/SOURCES.txt
+openenv_code_review.egg-info/dependency_links.txt
+openenv_code_review.egg-info/entry_points.txt
+openenv_code_review.egg-info/requires.txt
+openenv_code_review.egg-info/top_level.txt
+server/__init__.py
+server/app.py
+server/code_review_environment.py

openenv_code_review.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

openenv_code_review.egg-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+server = code_review.server.app:main

openenv_code_review.egg-info/requires.txt

@@ -0,0 +1,5 @@
+openenv-core[core]>=0.2.1
+
+[dev]
+pytest>=8.0.0
+pytest-cov>=4.0.0

openenv_code_review.egg-info/top_level.txt

@@ -0,0 +1 @@
+code_review

server/app.py

@@ -35,12 +35,12 @@ except Exception as e:  # pragma: no cover
         "openenv is required for the web interface. Install dependencies with '\n    uv sync\n'"
     ) from e
 
-# try:
-#     from ..models import CodeReviewAction, CodeReviewObservation
-#     from .code_review_environment import CodeReviewEnvironment
-# except ModuleNotFoundError:
-from models import CodeReviewAction, CodeReviewObservation
-from server.code_review_environment import CodeReviewEnvironment
+try:
+    from ..models import CodeReviewAction, CodeReviewObservation
+    from .code_review_environment import CodeReviewEnvironment
+except ModuleNotFoundError:
+    from models import CodeReviewAction, CodeReviewObservation
+    from server.code_review_environment import CodeReviewEnvironment
 
 
 # Create the app with web interface and README integration

server/code_review_environment.py

@@ -242,15 +242,15 @@ class CodeReviewEnvironment(Environment):
     def grade_action(self, action, ground_truth):
         score = 0.0
 
-        print("Action === ", action)
-        print("Ground truth === ", ground_truth)
+        # print("Action === ", action)
+        # print("Ground truth === ", ground_truth)
 
         # ------------------------------
         # ISSUE DETECTION (40%)
         # ------------------------------
         issue_score = self.score_issues(action.comment, ground_truth)
         score += 0.4 * issue_score
-        print("After Issue Score == ", issue_score)
+        # print("After Issue Score == ", issue_score)
 
         # ------------------------------
         # FIX QUALITY (30%)
@@ -258,7 +258,7 @@ class CodeReviewEnvironment(Environment):
         fix_score = self.score_fix(action.suggested_code, ground_truth)
         score += 0.3 * fix_score
 
-        print("After Fix Score == ", fix_score)
+        # print("After Fix Score == ", fix_score)
 
         # ------------------------------
         # DECISION (30%)
@@ -266,7 +266,7 @@ class CodeReviewEnvironment(Environment):
         decision_score = self.score_decision(action, ground_truth)
         score += 0.3 * decision_score
 
-        print("After Decision Score == ", decision_score)
+        # print("After Decision Score == ", decision_score)
 
         # ------------------------------
         # CLAMP SCORE