Upload folder using huggingface_hub

README.md

@@ -0,0 +1,26 @@
+---
+tags:
+- reinforcement-learning
+- openenv
+- environment
+---
+
+# Number Guessing Game Environment
+
+A simple RL environment where an agent tries to guess a randomly chosen number between 1 and 100.
+
+## Action
+
+- `guess` (int): A number between 1-100
+
+## Observation
+
+- `last_guess_feedback` (str): "Too high!", "Too low!", or "Correct!"
+- `guesses_remaining` (int): How many guesses are left
+- `reward` (float): +10 for correct, -1 for wrong
+- `done` (bool): Whether the episode is over
+
+## Rules
+
+- The agent has 5 guesses to find the number
+- Optimal strategy is binary search (can solve in ~7 guesses worst case)

models.py

@@ -0,0 +1,11 @@
+from openenv.core.env_server.types import Action, Observation
+
+
+class GuessAction(Action):
+    guess: int
+
+class GuessObservation(Observation):
+    last_guess_feedback: str
+    guesses_remaining: int
+    reward: float
+    done: bool

server/app.py

@@ -0,0 +1,20 @@
+from fastapi import FastAPI
+from pydantic import BaseModel
+
+from guess_env.server.guess_environment import GuessEnvironment
+from guess_env.models import GuessAction
+
+app = FastAPI()
+env = GuessEnvironment(max_guesses=5)
+
+class StepRequest(BaseModel):
+    action: GuessAction
+
+@app.post("/reset")
+def reset():
+    return env.reset()
+
+@app.post("/step")
+def step(request: StepRequest):
+    obs = env.step(request.action)
+    return obs

server/guess_environment.py

@@ -0,0 +1,48 @@
+from guess_env.models import GuessObservation
+import random
+
+class GuessEnvironment:
+    def __init__(self, max_guesses: int):
+        self.max_guesses = max_guesses
+        self.guesses_made = 0
+        self.done = False
+
+    def reset(self):
+        self.guesses_made = 0
+        self.done = False
+        self.target_number = random.randint(1, 100)
+        return GuessObservation(
+            last_guess_feedback="Game reset. Start guessing!",
+            guesses_remaining=self.max_guesses,
+            reward=0.0,
+            done=self.done
+        )
+    
+    def step(self, action):
+        if self.done:
+            raise Exception("Game is over. Please reset the environment.")
+
+        self.guesses_made += 1
+        guess = action.guess
+
+        if guess < self.target_number:
+            feedback = "Too low!"
+            reward = -1.0
+        elif guess > self.target_number:
+            feedback = "Too high!"
+            reward = -1.0
+        else:
+            feedback = "Correct! You've guessed the number!"
+            reward = 10.0
+            self.done = True
+
+        if self.guesses_made >= self.max_guesses and not self.done:
+            feedback = f"Game over! The correct number was {self.target_number}."
+            self.done = True
+
+        return GuessObservation(
+            last_guess_feedback=feedback,
+            guesses_remaining=self.max_guesses - self.guesses_made,
+            reward=reward,
+            done=self.done
+        )