Commit ·
75ec6ba
1
Parent(s): 1702cd1
final (#2)
Browse files- final (7142446df297e3d9557c4a2b92ea474c1943795d)
Co-authored-by: Vansh Bhardwaj <VanshBh@users.noreply.huggingface.co>
- server/app.py +26 -33
server/app.py
CHANGED
|
@@ -8,28 +8,23 @@ from environment import SupportEnv
|
|
| 8 |
|
| 9 |
app = FastAPI()
|
| 10 |
|
| 11 |
-
|
| 12 |
-
# Models
|
| 13 |
class Action(BaseModel):
|
| 14 |
message: str
|
| 15 |
|
| 16 |
-
|
| 17 |
class ResetRequest(BaseModel):
|
| 18 |
task: str = "easy"
|
| 19 |
|
| 20 |
-
|
| 21 |
class Observation(BaseModel):
|
| 22 |
echoed_message: str
|
| 23 |
|
| 24 |
-
|
| 25 |
class StepResponse(BaseModel):
|
| 26 |
observation: Observation
|
| 27 |
reward: float
|
| 28 |
done: bool
|
| 29 |
-
|
| 30 |
|
| 31 |
-
|
| 32 |
-
# Parser
|
| 33 |
def parse_action(message):
|
| 34 |
message = message.lower()
|
| 35 |
|
|
@@ -44,33 +39,31 @@ def parse_action(message):
|
|
| 44 |
else:
|
| 45 |
return "reject"
|
| 46 |
|
| 47 |
-
|
| 48 |
-
# TASKS
|
| 49 |
TASKS = {"easy": easy_cases, "medium": medium_cases, "hard": hard_cases}
|
| 50 |
|
| 51 |
-
# AUTO TASK CYCLING
|
| 52 |
-
TASK_SEQUENCE = ["easy", "medium", "hard"]
|
| 53 |
-
current_task_index = 0
|
| 54 |
-
|
| 55 |
env = None
|
| 56 |
|
|
|
|
| 57 |
|
| 58 |
# RESET
|
| 59 |
@app.post("/reset")
|
| 60 |
def reset(req: Optional[ResetRequest] = None):
|
| 61 |
-
global env
|
| 62 |
-
|
| 63 |
-
|
| 64 |
-
task = TASK_SEQUENCE[current_task_index]
|
| 65 |
-
current_task_index = (current_task_index + 1) % 3
|
| 66 |
|
| 67 |
-
cases = TASKS[
|
| 68 |
env = SupportEnv(cases)
|
| 69 |
|
| 70 |
obs = env.reset()
|
| 71 |
|
| 72 |
-
return {
|
| 73 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 74 |
|
| 75 |
# STATE
|
| 76 |
@app.get("/state")
|
|
@@ -81,8 +74,7 @@ def get_state():
|
|
| 81 |
|
| 82 |
return {"observation": {"echoed_message": env._get_obs()}}
|
| 83 |
|
| 84 |
-
|
| 85 |
-
# STEP
|
| 86 |
@app.post("/step", response_model=StepResponse)
|
| 87 |
def step(action: Action):
|
| 88 |
global env
|
|
@@ -91,30 +83,31 @@ def step(action: Action):
|
|
| 91 |
if env is None:
|
| 92 |
return {
|
| 93 |
"observation": {"echoed_message": "Error: Environment not initialized."},
|
| 94 |
-
"reward": 0.
|
| 95 |
"done": True,
|
|
|
|
| 96 |
}
|
| 97 |
|
| 98 |
parsed_action = parse_action(action.message)
|
| 99 |
|
| 100 |
obs, reward, done = env.step(parsed_action)
|
| 101 |
|
| 102 |
-
response = {
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 103 |
|
| 104 |
-
# RETURN SCORE PER TASK
|
| 105 |
if done:
|
| 106 |
-
response["score"] = env.get_score()
|
| 107 |
|
| 108 |
return response
|
| 109 |
|
| 110 |
-
|
| 111 |
def main():
|
| 112 |
import uvicorn
|
| 113 |
-
|
| 114 |
uvicorn.run(app, host="0.0.0.0", port=7860)
|
| 115 |
|
| 116 |
-
|
| 117 |
-
if __name__ == "__main__":
|
| 118 |
-
main()
|
| 119 |
if __name__ == "__main__":
|
| 120 |
main()
|
|
|
|
| 8 |
|
| 9 |
app = FastAPI()
|
| 10 |
|
| 11 |
+
# --- Models ---
|
|
|
|
| 12 |
class Action(BaseModel):
|
| 13 |
message: str
|
| 14 |
|
|
|
|
| 15 |
class ResetRequest(BaseModel):
|
| 16 |
task: str = "easy"
|
| 17 |
|
|
|
|
| 18 |
class Observation(BaseModel):
|
| 19 |
echoed_message: str
|
| 20 |
|
|
|
|
| 21 |
class StepResponse(BaseModel):
|
| 22 |
observation: Observation
|
| 23 |
reward: float
|
| 24 |
done: bool
|
| 25 |
+
info: dict = {}
|
| 26 |
|
| 27 |
+
# --- Parser ---
|
|
|
|
| 28 |
def parse_action(message):
|
| 29 |
message = message.lower()
|
| 30 |
|
|
|
|
| 39 |
else:
|
| 40 |
return "reject"
|
| 41 |
|
| 42 |
+
# --- TASKS ---
|
|
|
|
| 43 |
TASKS = {"easy": easy_cases, "medium": medium_cases, "hard": hard_cases}
|
| 44 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 45 |
env = None
|
| 46 |
|
| 47 |
+
# --- Routes ---
|
| 48 |
|
| 49 |
# RESET
|
| 50 |
@app.post("/reset")
|
| 51 |
def reset(req: Optional[ResetRequest] = None):
|
| 52 |
+
global env
|
| 53 |
+
|
| 54 |
+
task_name = req.task if req and req.task in TASKS else "easy"
|
|
|
|
|
|
|
| 55 |
|
| 56 |
+
cases = TASKS[task_name]
|
| 57 |
env = SupportEnv(cases)
|
| 58 |
|
| 59 |
obs = env.reset()
|
| 60 |
|
| 61 |
+
return {
|
| 62 |
+
"observation": {"echoed_message": obs},
|
| 63 |
+
"reward": 0.0,
|
| 64 |
+
"done": False,
|
| 65 |
+
"info": {}
|
| 66 |
+
}
|
| 67 |
|
| 68 |
# STATE
|
| 69 |
@app.get("/state")
|
|
|
|
| 74 |
|
| 75 |
return {"observation": {"echoed_message": env._get_obs()}}
|
| 76 |
|
| 77 |
+
# STEP
|
|
|
|
| 78 |
@app.post("/step", response_model=StepResponse)
|
| 79 |
def step(action: Action):
|
| 80 |
global env
|
|
|
|
| 83 |
if env is None:
|
| 84 |
return {
|
| 85 |
"observation": {"echoed_message": "Error: Environment not initialized."},
|
| 86 |
+
"reward": 0.0,
|
| 87 |
"done": True,
|
| 88 |
+
"info": {}
|
| 89 |
}
|
| 90 |
|
| 91 |
parsed_action = parse_action(action.message)
|
| 92 |
|
| 93 |
obs, reward, done = env.step(parsed_action)
|
| 94 |
|
| 95 |
+
response = {
|
| 96 |
+
"observation": {"echoed_message": obs},
|
| 97 |
+
"reward": reward,
|
| 98 |
+
"done": done,
|
| 99 |
+
"info": {}
|
| 100 |
+
}
|
| 101 |
|
|
|
|
| 102 |
if done:
|
| 103 |
+
response["info"]["score"] = float(env.get_score())
|
| 104 |
|
| 105 |
return response
|
| 106 |
|
| 107 |
+
# --- Main ---
|
| 108 |
def main():
|
| 109 |
import uvicorn
|
|
|
|
| 110 |
uvicorn.run(app, host="0.0.0.0", port=7860)
|
| 111 |
|
|
|
|
|
|
|
|
|
|
| 112 |
if __name__ == "__main__":
|
| 113 |
main()
|