Nebius Fine-Tuning

Fine-tune foundation models (LoRA or full weights) on Nebius Token Factory using the OpenAI-compatible API.

Prerequisites

pip install openai
export NEBIUS_API_KEY="your-key"

API base: https://api.tokenfactory.nebius.com/v1/

Supported base models

Key deployable models:

• meta-llama/Meta-Llama-3.1-8B-Instruct — LoRA + full fine-tuning, LoRA deployable serverlessly
• meta-llama/Llama-3.3-70B-Instruct — LoRA fine-tuning, LoRA deployable serverlessly