Add ndarray-cache.json (required by web-llm)

Update README with v4b benchmark results

Update README with v3 training results (76% accuracy, 64% NV recall)

Upload folder using huggingface_hub

Update README with v2 training data, benchmark results (81%/85%)

Upload folder using huggingface_hub

Upload README.md with huggingface_hub

Upload folder using huggingface_hub

Add model card with training details and usage examples

Upload folder using huggingface_hub

initial commit