> ## Documentation Index
> Fetch the complete documentation index at: https://docs.fireworks.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Reinforcement Fine-Tuning
> Train models using reinforcement learning in minutes
Fireworks RFT helps you train frontier models like DeepSeek V3 and Kimi K2 to **outperform closed models for your product use case, using reinforcement learning.** Fireworks RFT is powerful and easy to use for developers and enterprises:
* **No infrastructure:** Train frontier models without managing GPUs or RL infra
* **Production-ready:** Built-in tracing, monitoring, security & one-click deploy
* **Fast iteration:** From evaluator setup to deployed model in hours, not weeks
See how [Genspark](https://fireworks.ai/blog/genspark) and [Vercel](https://fireworks.ai/blog/vercel) used Fireworks RFT to train open models for agentic use cases, outperforming leading closed models.
## Quickstart: pick your training approach
**15 minutes** — test locally with a simple evaluator and small model.
**1–2 hours** — multi-turn rollouts in your environment with HTTP tracing.
**2–4 hours** — datasets stay in your GCS/S3 bucket.
## How RFT works
In supervised fine-tuning you provide labeled examples of good outputs. In RFT you provide **prompts** and an **evaluator** (reward function) that scores model outputs from 0 to 1. Training iteratively improves outputs that maximize that score.
Define how to score outputs (rules, tests, LLM-as-judge, or hybrid).
JSONL with prompts (`messages` array). Ground-truth completions are optional.
Local evaluators for simple tasks, or [remote environments](/fine-tuning/connect-environments) for agents.
Fireworks UI, `firectl rftj`, or [eval-protocol CLI](https://evalprotocol.io/introduction).
Promote the checkpoint and deploy LoRA on [on-demand hardware](/fine-tuning/deploying-loras).
RFT fits when the task is **verifiable** (you can judge output quality), you lack golden completions, or the task needs multi-step reasoning.
## Launch and operate
Use the docs for path choice and quickstarts. Use the **[Fireworks training skill](https://github.com/fw-ai/cookbook/tree/main/skills/fireworks-training)** for launch flags, validation, monitoring, and troubleshooting.
| Topic | Skill reference |
| ---------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------- |
| Prerequisites, validation, job states, CLI/UI launch | [managed-rft-operations](https://github.com/fw-ai/cookbook/blob/main/skills/fireworks-training/references/managed-rft-operations.md) |
| Evaluator design and preference data | [preference-data-and-evaluators](https://github.com/fw-ai/cookbook/blob/main/skills/fireworks-training/references/preference-data-and-evaluators.md) |
| Remote agent tracing | [rft-agent-tracing](https://github.com/fw-ai/cookbook/blob/main/skills/fireworks-training/references/rft-agent-tracing.md) |
| Parameter tuning and method choice | [choose-method](https://github.com/fw-ai/cookbook/blob/main/skills/fireworks-training/references/choose-method.md) |
| Failures and stuck jobs | [error-reference](https://github.com/fw-ai/cookbook/blob/main/skills/fireworks-training/references/error-reference.md) |
| Deploy and prove serving | [deploy-and-troubleshoot](https://github.com/fw-ai/cookbook/blob/main/skills/fireworks-training/references/deploy-and-troubleshoot.md) |
Prerequisites
Confirm account access and quota, an RFT-compatible base model, valid prompt JSONL, and a reviewed evaluator that produces non-flat scores on a small probe.
Evaluators
Evaluators turn a rollout into a reward. Test full-credit, partial-credit, zero, malformed-output, and edge cases locally before registration. Keep evaluator and preference-data guidance in the [evaluator skill reference](https://github.com/fw-ai/cookbook/blob/main/skills/fireworks-training/references/preference-data-and-evaluators.md).
Parameter tuning
Resolve the current defaults from `firectl rftj create --help` and the live RFT parameter reference. Start with a small run, change one optimization or rollout variable at a time, and compare reward and held-out quality.
Weighted training
Use per-example or per-token weighting only when the selected managed contract supports it. Inspect the rendered samples and effective loss mask before launch.
Launch from the CLI
Use `firectl rftj create` for a reproducible managed job. Prepare the full base-model, dataset, evaluator, and output-model resource names, run the documented preflight, and confirm the resolved plan before creation.
Launch from the dashboard
Use the dashboard for a guided first run: choose the model, dataset, evaluator, rollout settings, and optional W\&B project, then review the summary before launch.
Warm start
Warm-start from a compatible promoted LoRA when continuing SFT into RFT. Do not also supply a conflicting base-model field; verify the current CLI contract before creation.
Evaluator secrets
Store credentials in the Fireworks secret-management surface and reference them by environment-variable name. Never embed secret values in evaluator code, datasets, or run reports.
Cost planning
Estimate rollout inference, training runtime or tokens, candidate count, response length, and any deployment uptime. Use the live pricing page and [Price vs Tinker](/fine-tuning/multi-turn-cost-comparison); do not assume a free tier.
Monitor training
Track state plus actual progress: optimizer steps, rollout throughput, reward distribution, validation results, logs, and W\&B metrics when enabled. Inspect representative rollouts for reward hacking and stop when quality regresses or the evaluator saturates.
Using Cursor, Claude Code, or Codex? [Install Agent Skills](/fine-tuning/agent/use-with-coding-agents) to drive managed RFT with the `fireworks-training` skill.
Parameter field reference: [RFT parameters](/fine-tuning/rft-parameters-reference) (API tab).
## Related
HTTP `/init` contract for remote rollouts
Multi-turn RL cost comparison
Custom RL loops on dedicated or serverless infrastructure
Which models support managed RFT