igalk474/MiniFrontier-150M-Modern-3B-token-mixture

MiniFrontier 150M-Modern 5B-token mixture Training-mixture export from MiniFrontier - AI-LLM-Transformers-Edu-Model, an educational+modern, from-scratch decoder-only language model. Each row is one admitted document (post-filter, post-dedup, pre-tokenization) with its full provenance: text, source, revision, license, language, record_id, content_hash, path, source_type, split, parent_content_hash, transform. License is per-example, not one blanket license for the dataset. The… See the full description on the dataset page: https://huggingface.co/datasets/igalk474/MiniFrontier-150M-Modern-3B-token-mixture.