Kraken is GENOMA Labs' curated agentic + coding knowledge foundation for the KALYPSO model line. This public release is ToS-clean by construction and decontaminated against the standard coding benchmarks.
Dataset Details
Description
Kraken-Public v1 is an 18,049-record agentic-and-coding supervised-fine-tuning corpus, curated by GENOMA Labs from CC-BY-4.0 source data and decontaminated (13-gram overlap removal)… See the full description on the dataset page: https://huggingface.co/datasets/GenomaLabs-com/Kraken.
