Buckets:
| <meta charset="utf-8" /><meta name="hf:doc:metadata" content="{"title":"Training with OpenEnv","local":"training-with-openenv","sections":[{"title":"Ways to Train","local":"ways-to-train","sections":[],"depth":2},{"title":"Integrations","local":"integrations","sections":[],"depth":2},{"title":"Your Own Training Loop","local":"your-own-training-loop","sections":[],"depth":2}],"depth":1}"/> | |
| <link href="/docs/openenv/main/en/_app/immutable/entry/start.BVsaIFPz.js" rel="modulepreload"> | |
| <link href="/docs/openenv/main/en/_app/immutable/chunks/DOr2Gjb3.js" rel="modulepreload"> | |
| <link href="/docs/openenv/main/en/_app/immutable/chunks/DfHjNWj2.js" rel="modulepreload"> | |
| <link href="/docs/openenv/main/en/_app/immutable/entry/app.Bi0HOH_m.js" rel="modulepreload"> | |
| <link href="/docs/openenv/main/en/_app/immutable/chunks/Ba1drWuH.js" rel="modulepreload"> | |
| <link href="/docs/openenv/main/en/_app/immutable/chunks/Cs50_Xpk.js" rel="modulepreload"> | |
| <link href="/docs/openenv/main/en/_app/immutable/chunks/DsnmJJEf.js" rel="modulepreload"> | |
| <link href="/docs/openenv/main/en/_app/immutable/chunks/PGLvW8OU.js" rel="modulepreload"> | |
| <link href="/docs/openenv/main/en/_app/immutable/nodes/0.uj0bKvXN.js" rel="modulepreload"> | |
| <link href="/docs/openenv/main/en/_app/immutable/chunks/CFswWuIO.js" rel="modulepreload"> | |
| <link href="/docs/openenv/main/en/_app/immutable/nodes/13.Hb6_LZqP.js" rel="modulepreload"> | |
| <link href="/docs/openenv/main/en/_app/immutable/chunks/DCZF09PK.js" rel="modulepreload"> | |
| <!--1mbfh0f--><meta name="hf:doc:metadata" content="{"title":"Training with OpenEnv","local":"training-with-openenv","sections":[{"title":"Ways to Train","local":"ways-to-train","sections":[],"depth":2},{"title":"Integrations","local":"integrations","sections":[],"depth":2},{"title":"Your Own Training Loop","local":"your-own-training-loop","sections":[],"depth":2}],"depth":1}"/><!----> | |
| <link href="/docs/openenv/main/en/_app/immutable/assets/0.tn0RQdqM.css" rel="modulepreload"> <!--[--><!--[0--><!--[--><!--[0--><!--[--><p></p> <div class="items-center shrink-0 min-w-[100px] max-sm:min-w-[50px] justify-end ml-auto flex" style="float: right; margin-left: 10px; display: inline-flex; position: relative; z-index: 10;"><div class="inline-flex rounded-md max-sm:rounded-sm"><button class="inline-flex items-center gap-1 h-7 max-sm:h-7 px-2 max-sm:px-1.5 text-sm font-medium text-gray-800 border border-r-0 rounded-l-md max-sm:rounded-l-sm border-gray-200 bg-white hover:shadow-inner dark:border-gray-850 dark:bg-gray-950 dark:text-gray-200 dark:hover:bg-gray-800" aria-live="polite"><span class="inline-flex items-center justify-center rounded-md p-0.5 max-sm:p-0 hover:text-gray-800 dark:hover:text-gray-200"><svg class="sm:size-3.5 size-3" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----></span> <span>Copy page</span></button> <button class="inline-flex items-center justify-center w-6 max-sm:w-5 h-7 max-sm:h-7 disabled:pointer-events-none text-sm text-gray-500 hover:text-gray-700 dark:hover:text-white rounded-r-md max-sm:rounded-r-sm border border-l transition border-gray-200 bg-white hover:shadow-inner dark:border-gray-850 dark:bg-gray-950 dark:text-gray-200 dark:hover:bg-gray-800" aria-haspopup="menu" aria-expanded="false" aria-label="Open copy menu"><svg class="transition-transform text-gray-400 overflow-visible sm:size-3.5 size-3 rotate-0" width="1em" height="1em" viewBox="0 0 12 7" fill="none" xmlns="http://www.w3.org/2000/svg"><path d="M1 1L6 6L11 1" stroke="currentColor"></path></svg><!----></button></div> <!--[-1--><!--]--></div><!----> <!--[0--><h1 class="relative group"><a id="training-with-openenv" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#training-with-openenv"><span><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg><!----></span></a> <span>Training with OpenEnv</span></h1><!--]--><!----> <p>OpenEnv defines the contract between environments and the code that uses them. Training stays in your framework. An environment exposes <code>reset()</code>, <code>step()</code> and <code>state()</code> (and its tools over MCP), runs as a service anywhere, and computes its own reward, so any training framework that can call it can train on it. This page maps the ways to train on OpenEnv environments and the frameworks that support it.</p> <!--[1--><h2 class="relative group"><a id="ways-to-train" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#ways-to-train"><span><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg><!----></span></a> <span>Ways to Train</span></h2><!--]--><!----> <table><thead><tr><th>You want to</th><th>What the environment gives</th><th>Worked examples</th></tr></thead><tbody><tr><td>Train a model with RL, with the trainer running the episode</td><td>Observations, tools and a reward. The trainer generates every turn, so it has the tokens and logprobs. This is the white-box path</td><td><a href="../tutorials/wordle-grpo">Wordle</a>, <a href="../tutorials/rl-training-2048">2048</a>, <a href="../tutorials/end-to-end-walkthrough">a reasoning model</a> and <a href="../tutorials/browsergym-harness">web tasks in BrowserGym</a> with TRL, <a href="https://colab.research.google.com/github/unslothai/notebooks/blob/main/nb/OpenEnv_gpt_oss_(20B)_Reinforcement_Learning_2048_Game.ipynb" rel="nofollow">2048 with gpt-oss</a> with Unsloth, and the other frameworks’ examples under <a href="#integrations">Integrations</a></td></tr><tr><td>Train the model behind a real agent (OpenCode, Claude Code, Codex, …)</td><td>Harbor runs the agent on a task, captures every model call, and returns a <code>TrainingTrace</code> with token ids, logprobs and the verifier’s reward. This is the black-box path</td><td><a href="../tutorials/harbor-harness">Harbor</a>, trained with TRL’s <code>AsyncGRPOTrainer</code></td></tr><tr><td>Warm-start a model with supervised data</td><td><code>openenv collect</code> runs a teacher in the environment and saves reward-labeled rollouts as a dataset</td><td><a href="../tutorials/sft-warmup">Collecting rollouts for SFT</a></td></tr><tr><td>Evaluate, not train</td><td>Scores from the environment’s rubric</td><td><a href="../tutorials/evaluation-inspect">Inspect AI</a>, <a href="../tutorials/claude-code-harness">an agent inside the environment</a></td></tr></tbody></table> <p>None of these paths ties the environment to a framework. <code>openenv.core</code> does not depend on any trainer, and a <code>TrainingTrace</code> or a collected dataset is plain data that any trainer can read. The worked examples use the frameworks named above because that is where the examples exist today. <a href="../tutorials/harnesses">Harnesses in OpenEnv</a> compares the white-box and black-box paths in more detail.</p> <!--[1--><h2 class="relative group"><a id="integrations" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#integrations"><span><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg><!----></span></a> <span>Integrations</span></h2><!--]--><!----> <p>These frameworks and platforms train on OpenEnv environments. If your project supports OpenEnv, open a PR to add it here and to the <a href="https://github.com/huggingface/OpenEnv#integrations" rel="nofollow">README</a>.</p> <table><thead><tr><th>Framework</th><th>Example</th></tr></thead><tbody><tr><td><a href="https://art.openpipe.ai" rel="nofollow">ART</a></td><td><a href="https://art.openpipe.ai/integrations/openenv-integration" rel="nofollow">OpenEnv integration</a></td></tr><tr><td><a href="https://lightning.ai" rel="nofollow">Lightning AI</a></td><td><a href="https://lightning.ai/templates?section=featured&query=openenv" rel="nofollow">OpenEnv templates</a></td></tr><tr><td><a href="https://github.com/radixark/miles" rel="nofollow">Miles</a></td><td><a href="https://github.com/radixark/miles/tree/main/examples/experimental/openenv" rel="nofollow">GRPO on Terminal-Bench 2</a></td></tr><tr><td><a href="https://github.com/oumi-ai/oumi" rel="nofollow">Oumi</a></td><td><a href="https://github.com/oumi-ai/oumi/blob/main/notebooks/Oumi%20-%20OpenEnv%20GRPO%20with%20trl.ipynb" rel="nofollow">OpenEnv GRPO notebook</a></td></tr><tr><td><a href="https://github.com/NovaSky-AI/SkyRL" rel="nofollow">SkyRL</a></td><td><a href="https://skyrl.readthedocs.io/en/latest/examples/openenv.html" rel="nofollow">Training on OpenEnv environments</a></td></tr><tr><td><a href="https://meta-pytorch.org/torchforge/" rel="nofollow">torchforge</a></td><td><a href="https://github.com/huggingface/OpenEnv/tree/main/examples/grpo_blackjack" rel="nofollow">GRPO on BlackJack</a></td></tr><tr><td><a href="https://huggingface.co/docs/trl" rel="nofollow">TRL</a></td><td><a href="https://huggingface.co/docs/trl/openenv" rel="nofollow">OpenEnv integration guide</a>: <code>environment_factory</code> with <code>GRPOTrainer</code>, several environments at once, and harness training through Harbor</td></tr><tr><td><a href="https://unsloth.ai" rel="nofollow">Unsloth</a></td><td><a href="https://colab.research.google.com/github/unslothai/notebooks/blob/main/nb/OpenEnv_gpt_oss_(20B)_Reinforcement_Learning_2048_Game.ipynb" rel="nofollow">2048 with gpt-oss</a></td></tr></tbody></table> <!--[1--><h2 class="relative group"><a id="your-own-training-loop" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#your-own-training-loop"><span><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg><!----></span></a> <span>Your Own Training Loop</span></h2><!--]--><!----> <p>Every integration above comes down to the same calls. To plug OpenEnv into a framework that has no integration yet, drive the client from its rollout code:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copied</div><!----></button><!----></div> <pre class="language-python "><!----><span class="hljs-keyword">from</span> openenv <span class="hljs-keyword">import</span> AutoAction, AutoEnv | |
| env = AutoEnv.from_env(<span class="hljs-string">"my-env"</span>) | |
| Action = AutoAction.from_env(<span class="hljs-string">"my-env"</span>) <span class="hljs-comment"># the environment's action class</span> | |
| <span class="hljs-keyword">with</span> env.sync() <span class="hljs-keyword">as</span> client: | |
| <span class="hljs-keyword">for</span> episode <span class="hljs-keyword">in</span> <span class="hljs-built_in">range</span>(num_episodes): | |
| result = client.reset() | |
| <span class="hljs-keyword">while</span> <span class="hljs-keyword">not</span> result.done: | |
| action = Action(**policy(result.observation)) <span class="hljs-comment"># your model picks the next action</span> | |
| result = client.step(action) | |
| <span class="hljs-comment"># result.reward is the environment's reward for this step</span><!----></pre></div><!----> <p>The <a href="task-api">Task API</a> lets a trainer list an environment’s tasks and pick which one each episode runs, and <a href="rewards">Rewards</a> covers how environments compute the reward.</p> <a class="!text-gray-400 !no-underline text-sm flex items-center not-prose mt-4" href="https://github.com/huggingface/openenv/blob/main/docs/source/guides/training.md" target="_blank"><svg class="mr-1" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M31,16l-7,7l-1.41-1.41L28.17,16l-5.58-5.59L24,9l7,7z"></path><path d="M1,16l7-7l1.41,1.41L3.83,16l5.58,5.59L8,23l-7-7z"></path><path d="M12.419,25.484L17.639,6.552l1.932,0.518L14.351,26.002z"></path></svg><!----> <span><span class="underline">Update</span> on GitHub</span></a><!----> <p></p><!--]--><!----><!--]--><!--]--><!--]--> <!--[-1--><!--]--><!--]--> | |
| <script> | |
| { | |
| __sveltekit_8eecdv = { | |
| base: "/docs/openenv/main/en", | |
| assets: "/docs/openenv/main/en" | |
| }; | |
| const element = document.currentScript.parentElement; | |
| Promise.all([ | |
| import("/docs/openenv/main/en/_app/immutable/entry/start.BVsaIFPz.js"), | |
| import("/docs/openenv/main/en/_app/immutable/entry/app.Bi0HOH_m.js") | |
| ]).then(([kit, app]) => { | |
| kit.start(app, element, { | |
| node_ids: [0, 13], | |
| data: [null,null], | |
| form: null, | |
| error: null | |
| }); | |
| }); | |
| } | |
| </script> | |
Xet Storage Details
- Size:
- 17.5 kB
- Xet hash:
- fecfa86f52461169992f8edebb640a1b99f244ff332c822ac718af97a9d0b169
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.