docs: Added tutorial regarding RLTrainer

[onel]

Description

As per the comment in the issue, I create a plan about how to use RLTrainer, end to end.
From prerequisites, setting up, running, analyzing.

Issue #748

Type of Change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update
• Test improvement

Testing

• All existing tests pass when running uv run pytest locally.
• New tests have been added to cover the changes

Checklist

• My code follows the style guidelines of this project as outlined in AGENTS.md
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• Any dependent changes have been merged and published

Additional Notes

Used an AI agent to write the content but checked and corrected the final output.

---

Note

Low Risk
Documentation-only addition with no runtime or API changes; main risk is potential user confusion if commands/config examples drift from current tooling.

Overview
Adds a new end-to-end documentation page, docs/rltrainer-tutorial.md, that walks users through running an RL training job with vf.RLTrainer (workspace setup via prime lab setup --vf-rl, configuring a sample reverse-text TOML, launching training, and understanding outputs).

Includes guidance on interpreting metrics/checkpoints plus common troubleshooting and advanced configuration knobs (batch sizing, generation params, LoRA vs full finetune, wandb logging), along with next-step links to other environments and prime-rl.

^{Written by Cursor Bugbot for commit ce6879b. This will update automatically on new commits. Configure here.}

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you all sign our Contributor License Agreement before we can accept your contribution.
0 out of 2 committers have signed the CLA.

❌ onel
❌ askmanu[bot]
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[willccbb]

Hmmm I'm probably not gonna add this, sorry -- I don't really want to encourage people to use RLTrainer lol, it's unmaintained + not performance-optimized + not adding new features. Prime-rl should be used instead. It has all the same features, much better performance, and is heavily battle-tested + supports similar configuration. I'm keeping RLTrainer in the repo because some people already do use it + like it (much to my chagrin), and it's a convenient example for reading simple RL code for instructional purposes. But if someone wants to run a trainer as-is without modifying or reading the code, they should not use or think about RLTrainer at all.

[onel]

Hmmm I'm probably not gonna add this, sorry -- I don't really want to encourage people to use RLTrainer lol, it's unmaintained + not performance-optimized + not adding new features. Prime-rl should be used instead. It has all the same features, much better performance, and is heavily battle-tested + supports similar configuration. I'm keeping RLTrainer in the repo because some people already do use it + like it (much to my chagrin), and it's a convenient example for reading simple RL code for instructional purposes. But if someone wants to run a trainer as-is without modifying or reading the code, they should not use or think about RLTrainer at all.

@ChrisDelClea this might be helpful to you 👆

@willccbb makes perfect sense. do you want me to look into adding a similar tutorial for Prime-rl? would that be helpful?

[willccbb]

You're welcome to propose any ideas! Though FYI we're also working on our own content here + might be easier to wait for that and make suggestions for changes as needed?

[onel]

Both sound good. I'll wait for you to create new content.

If you have a certain type of documentation that's not priority or the team doesn't have bandwidth LMK, I can help with that.
Let's move the discussion to #748

Closing this PR for now.