---
title: OpenReward Echo Env (TRL test fixture)
emoji: 🦜
colorFrom: blue
colorTo: purple
sdk: docker
app_port: 7860
pinned: false
license: apache-2.0
---

# OpenReward Echo Env

Minimal [Open Reward Standard](https://openrewardstandard.io) environment used as the test fixture for [`trl.experimental.openreward`](https://github.com/huggingface/trl/tree/main/trl/experimental/openreward).

The model is given a target string and must call `echo(text=...)` with exactly that string. Reward is `1.0` on match, `0.0` otherwise; the episode finishes on a correct echo.

Pure Python — no sandbox, no external state — so responses are deterministic and the env can run thousands of concurrent sessions on free-tier hardware.

## Use

```python
import os
os.environ["OPENREWARD_API_URL"]     = "https://trl-internal-testing-openreward-echo-env.hf.space"
os.environ["OPENREWARD_SESSION_URL"] = "https://trl-internal-testing-openreward-echo-env.hf.space"

from trl.experimental.openreward import OpenRewardSpec

spec = OpenRewardSpec(
    "https://trl-internal-testing-openreward-echo-env.hf.space",
    env_name="echoenvironment",
)
print(spec.train_dataset)  # 8 rows: target ∈ {"hello", "world", "trl", ...}
```

The two `OPENREWARD_*_URL` overrides are needed because the `openreward` SDK by default expects a two-subdomain platform layout (`api.<host>` + `sessions.<host>`); for a single-host self-hosted server both have to point at the same URL.

## Tasks

| split | count | shape |
|---|---|---|
| `train` | 8 | `{"id": "echo-N", "target": "<word>"}` |

## Local development

```bash
pip install -r requirements.txt
python server.py    # listens on PORT (default 8080)
```

## Files

- `server.py` — env definition + `Server([EchoEnvironment]).run()`
- `Dockerfile` — `python:3.11-slim` + the deps; HF Spaces serves it on port 7860
- `requirements.txt` — `openreward`, `fastapi`, `uvicorn`