researchactiveSeptember 2024

Gemma Fine-tuning

Fine-tuning Gemma 3 4B for meeting intelligence

◆ What

Fine-tuning Gemma 3 4B for meeting intelligence

◆ How

Gemma 3QLoRATRLAxolotl

◆ Why

The Lab

Supporting: Solid production work

Overview

Fine-tuning Gemma 3 4B for meeting transcript synthesis and summarization. QLoRA training with TRL/Axolotl on DGX Spark. Related to Axon meeting intelligence system.

Technologies

Gemma 3QLoRATRLAxolotl

Narrative Framings

The Lab

Related Writing

Connected Projects

↔Related

personal

Axon

Executive second brain with RAG + agents

research

OncoVLM

Oncology multimodal AI models at three scales

Explore all projects