Skip to main content
Post-training in Nebius Token Factory lets you adapt foundation models to your data and tasks. By training on your own examples, the model learns your domain patterns directly improving accuracy, stability, and reducing prompt complexity. Supervised fine-tuning allows you to train on large datasets without prompt-length limits. This gives you tighter control over model behavior, reduces the need for manual prompt engineering, and can lead to lower inference cost and latency.

Upcoming Advanced post-training options including:
  • Speculative Decoding optimization (private beta [sign up])
  • Reinforcement Fine-tuning (limited professional service [request access])
You can currently run long-context supervised fine-tuning on full model weights or LoRA adapters.
If you’re ready to fine-tune or want to explore the workflows involved, start here:

How to Fine-tune

Step-by-step guide to preparing data, launching a job, and evaluating results.

Models

Supported base models for training and inference.

Datasets

Create and manage datasets for training and validation.

Deploy Custom LoRA

Serve LoRA-adapted models serverlessly with per-token billing.