Team Ai
Datasetpublic

xinbenlv/gemma-finetune-webgpu

gemma-finetune-webgpu Built voice/style instruction-tuned datasets used by the gemma-finetune workshop (May 2026, Immersive Commons). Each row is dolly-15k–shaped: {"instruction": "...", "context": "...", "response": "...", "category": "..."} Files file rows upstream recipe shakespeare_15k.jsonl 15,000 HF benchaffe/shakespeare-lines 12.5K 4-line continuation windows + 2.5K per-theme style obama_15k.jsonl 15,000 fivethirtyeight/data BarackObama.csv 6… See the full description on the dataset page: https://huggingface.co/datasets/xinbenlv/gemma-finetune-webgpu.

sourceHugging Facemitupdated 5mo agoView on Hugging Face
0likes29downloads
Dataset Card

gemma-finetune-webgpu

Built voice/style instruction-tuned datasets used by the gemma-finetune workshop (May 2026, Immersive Commons). Each row is dolly-15k–shaped:

json
{"instruction": "...", "context": "...", "response": "...", "category": "..."}

Files

filerowsupstreamrecipe
shakespeare_15k.jsonl15,000HF `benchaffe/shakespeare-lines`12.5K 4-line continuation windows + 2.5K per-theme style
obama_15k.jsonl15,000fivethirtyeight/data BarackObama.csv6 templates per cleaned tweet (style ×2, continuation, topic, tone, author classification)
trump_15k.jsonl15,000HF `fschlatt/trump-tweets`drop retweets + profanity-filter (default on); 7.5K style + 6.75K continuation + 750 author classification
marktwain_15k.jsonl15,000Project Gutenberg (10 books)10K continuation + 5K style passages

Reproducible from the build script in the upstream repo:

bash
git clone https://github.com/RayyanZahid/gemma-finetune
cd gemma-finetune
python data/build_voice_dataset.py --all

The build script uses random.Random(42); output is deterministic.

Use with Gemma fine-tuning

bash
python templates/finetune.py --user me --dataset data/obama_15k.jsonl --out-dir runs

License

MIT for the derived JSONL. Upstream sources: Project Gutenberg works (Twain) are public domain; fschlatt/trump-tweets is CC0; FiveThirtyEight tweet repo is open; benchaffe/shakespeare-lines is public domain Shakespeare.