Adapt Gemma 3n E4B for German audio transcription
An audio-input training example using Gemma 3n and a German speech dataset.
Published by Unsloth
How to use it
The notebook export loads Gemma 3n E4B and an Emilia German speech split, formats audio and text examples, and trains adapters for the model.
Run the publisher's notebook cells in the intended environment. For an independent check, keep speakers separate across train and evaluation splits and measure word error rate on held-out recordings.
- Framework
- Unsloth
- Dataset
- kadirnar/Emilia-DE-B000000
- Hardware
- The exported example uses CUDA. Minimum VRAM unspecified.
- Access
- Public publisher notebook export; model and dataset terms apply.