---
canonical: "https://www.softatlas.io/together-ai/products/custom-training"
name: "Custom Training"
provider: "Together AI"
category: "Artificial Intelligence / ML & Data Science Platforms / Data Science & ML Platforms"
review_count: 0
ai_powered: true
website: "https://together.ai/custom-training"
updated: "2026-10-05"
---

# Custom Training

By [Together AI](/together-ai.md)

Post-training for frontier quality at scale, allowing fine-tuning of open-source models for production workloads.

## Features
- Fine-tune open-source models with custom data
- Conduct reinforcement learning experiments at scale
- Run supervised fine-tuning with user-provided demonstrations
- Perform high-scale multi-LoRA inference
- Optimize models using reward and KL tracking
- Merge and quantize models for scalable serving
- Configure training jobs through a granular Python SDK
- Validate model quality with evaluations

## Built for
DEVELOPERS, OPERATIONS, EXECUTIVES

Source: https://www.softatlas.io/together-ai/products/custom-training