STaR × S1 math pipeline on Qwen2.5-1.5B. LoRA, strict Final: format, ~20–30% acc (OpenR1-Math split).
-
Updated
Sep 6, 2025 - Python
STaR × S1 math pipeline on Qwen2.5-1.5B. LoRA, strict Final: format, ~20–30% acc (OpenR1-Math split).
Examples demonstrating dataset api capabilities
Convert MIDI to ABC music files as a dataset pipeline for the Lakh MIDI Dataset, linked with the Million Song Dataset.
Computer vision pipeline for building annotated ML datasets from structural beam experiment photos — calibration, segmentation export, and perspective correction
Reproducible Spanish-news NLP pipeline: RSS collection → weak-supervised dataset → TF-IDF baselines → honest macro metrics. Runs end-to-end on a laptop in under a minute.
To associate your repository with the dataset-pipeline topic, visit your repo's landing page and select "manage topics."