Apppublic

prasanna2119/Customer_Support_RL_Environment

sourceHugging Faceupdated 5mo agoView on Hugging Face

0likes

App README

🎧 Customer Support RL Environment

An OpenEnv-compliant reinforcement learning environment that simulates a real-world customer support desk.

Overview

An AI agent learns to handle customer support workflows across three progressively harder tasks:

Task	Difficulty	Description
`classify-ticket`	Easy	Assign a support ticket to the correct category
`draft-response`	Medium	Write a professional empathetic response
`resolve-escalation`	Hard	Handle a multi-turn escalation sequence

Action Space

Each action is a JSON object:

json

{
  "action_type": "classify | draft | acknowledge_urgency | escalate_manager | ...",
  "content": "<text content of the action>"
}

Observation Space

Each observation is a JSON object containing:

—task — current task name
—session_id — unique episode identifier
—instruction — what the agent should do
—ticket or conversation — the support scenario
—action_format — expected format of the action

Reward Function

Task	Reward Signal
classify-ticket	1.0 correct, 0.5 related category, 0.0 wrong
draft-response	0.0–1.0 based on required elements present, forbidden elements absent, response length
resolve-escalation	1.0 correct action per step, 0.4 related action, 0.0 wrong

API Endpoints

Method	Path	Description
GET	`/health`	Health check
GET	`/tasks`	List all tasks
POST	`/reset`	Start a new episode
POST	`/step`	Take an action
GET/POST	`/state`	Query current episode state

Example: Reset

bash

curl -X POST http://localhost:7860/reset \
  -H "Content-Type: application/json" \
  -d '{"task": "classify-ticket"}'

Example: Step

bash

curl -X POST http://localhost:7860/step \
  -H "Content-Type: application/json" \
  -d '{"task": "classify-ticket", "action": {"action_type": "classify", "content": "billing"}}'

Setup & Running

Local

bash

pip install -r requirements.txt
uvicorn app:app --host 0.0.0.0 --port 7860

Docker

bash

docker build -t customer-support-env .
docker run -p 7860:7860 \
  -e HF_TOKEN=your_token \
  -e MODEL_NAME=Qwen/Qwen2.5-72B-Instruct \
  customer-support-env

Run Inference

bash

export HF_TOKEN=your_token
export MODEL_NAME=Qwen/Qwen2.5-72B-Instruct
export API_BASE_URL=https://router.huggingface.co/v1
export ENV_BASE_URL=http://localhost:7860

python inference.py

Environment Variables

Variable	Description	Default
`HF_TOKEN`	HuggingFace API key	—
`MODEL_NAME`	LLM model identifier	`Qwen/Qwen2.5-72B-Instruct`
`API_BASE_URL`	LLM API endpoint	`https://router.huggingface.co/v1`
`ENV_BASE_URL`	This environment's URL	`http://localhost:7860`
`TASK`	Run a specific task only	all tasks

Team

AgentForge — OpenEnv Hackathon 2026