Large language models (LLMs)
This feature is in Beta. Workspace admins can control access to this feature from the Previews page. See Manage Databricks previews.
This page provides notebook examples for fine-tuning large language models (LLMs) using Serverless GPU compute. These examples demonstrate various approaches to fine-tuning including parameter-efficient methods like Low-Rank Adaptation (LoRA) and full supervised fine-tuning.
Fine-tune Qwen2-0.5B model
The following notebook provides an example of how to efficiently fine-tune the Qwen2-0.5B model using:
- Transformer reinforcement learning (TRL) for supervised fine-tuning
- Liger Kernels for memory-efficient training with optimized Triton kernels.
- LoRA for parameter-efficient fine-tuning.
Notebook
Fine-tune Llama-3.2-3B with Unsloth
This notebook demonstrates how to fine-tune Llama-3.2-3B using the Unsloth library.
Unsloth Llama
Video demo
This video walks through the notebook in detail (12 minutes).
Fine-tune a GPT OSS 20B model
This notebook demonstrates how to fine-tune OpenAI's gpt-oss-20b model on a H100 GPU using LoRA for parameter-efficient fine-tuning.
Fine-tune GPT
Supervised fine-tuning using DeepSpeed and TRL
This notebook demonstrates how to use the Serverless GPU Python API to run supervised fine-tuning (SFT) using the Transformer Reinforcement Learning (TRL) library with DeepSpeed ZeRO Stage 3 optimization.
TRL DeepSpeed
LORA fine-tuning using Axolotl
This notebook demostrates how to use the Serverless GPU Python API to LORA fine-tune an Olmo3 7B model using the Axolotl library.