> ## Documentation Index
> Fetch the complete documentation index at: https://benchgen.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Train a model on benchmark results

> The skill that plans and launches fine-tuning, reinforcement learning and distillation, follows the job, and exports a merged model.

| The skill | `training-launch` |
| - | - |
| **Changes anything** | Yes. Training holds a GPU and spends credits, so it always plans first. |
| **It picks this up when you say** | "fine-tune this model", "train on what it got wrong", "distill it into a smaller model", "how is my training job", "merge the adapter" |
| **It hands over to** | `benchmark-launch` to measure whether the training actually helped |

## What it can do

* Plan and launch a LoRA fine-tune
* Plan and launch reinforcement learning with GRPO
* Continue training an existing adapter
* Train straight from a benchmark run
* Show what a job would submit before anything starts
* Report a job's status and tail its logs
* Show the dataset a finished run produced
* Merge a trained adapter and publish it as standalone weights
* Tell you which GPUs are free

Training from a run uses the dataset that run produced:

| Mode | What it trains on |
| - | - |
| `failures` | The questions the model got wrong, with the correct answers. The default. |
| `all` | Every question and answer. |
| `distill` | The questions a **strong** model got right, with its own reasoning. |

![The BenchGen AI agent's training plan: from-run distillation, base model Qwen2.5-0.5B-Instruct, source run 1081, dataset mode distill, 30 correct trajectories, 10 epochs, LoRA rank 16, and a note that nothing has been launched](https://mintcdn.com/benchgen-8fc81371/MiEv2r5AKjiUikhl/images/guides/benchgen-ai-agent/03-distillation-plan.png?fit=max\&auto=format\&n=MiEv2r5AKjiUikhl\&q=85\&s=192370956835ed53f0b997dd10fbbfb6)

A training plan: what would be submitted, and nothing launched until you confirm

## Prompts to try

```text theme={null}
Plan a LoRA fine-tune of Qwen/Qwen2.5-0.5B-Instruct on yahma/alpaca-cleaned, 1 epoch,
50 steps maximum.
```

```text theme={null}
Plan a training from run <id> so the model learns from the questions it got wrong.
```

```text theme={null}
Plan a distillation from run <id>, dataset mode distill, base model Qwen/Qwen2.5-0.5B-Instruct.
```

```text theme={null}
What GPUs are free right now? How is job <id> doing?
```

## What it will not do

* Start a job without a plan and a confirmation. Anything other than a yes cancels
* Run the whole loop unattended. Each step is confirmed on its own
* Stop or delete a training job. Do that from the web app
* Decide for you whether the model is the problem

> [!WARNING]
>
> The base model is a Hugging Face repository id such as `Qwen/Qwen2.5-0.5B-Instruct`, not a
> catalogue display name. Check the `base_model` line in the plan before confirming.

> [!NOTE]
>
> If the misses came from a broken answer key, [fix the benchmark](https://benchgen.com/docs/skills/edit-a-benchmark.md)
> first. Training against a wrong key teaches the wrong answer.

## Related

> [card] [The full improvement loop](https://benchgen.com/docs/guides/benchgen-ai-agent/working-with-the-agent.md#close-the-full-improvement-loop)
>
> Measure, train, measure again, including distillation with real numbers.

> [card] [Analyze the results](https://benchgen.com/docs/skills/analyze-results.md)
>
> Decide whether the misses are the model's fault first.

> [card] [Fine-tune a model](https://benchgen.com/docs/train/fine-tune-a-model.md)
>
> The same training from the web app.

> [card] [Launch training (API)](https://benchgen.com/docs/api-reference/endpoint/train.md)
>
> Start jobs from your own code.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.