Pinned Loading

1. evaluation-harness evaluation-harness Public

   A from-scratch evaluation and benchmarking framework for language models and agents: pluggable tasks and graders, bootstrap confidence intervals, significance testing, and contamination checks.

   Python

2. ann-vector-search-library ann-vector-search-library Public

   An approximate nearest neighbor vector search library built from scratch in pure Python, implementing HNSW graphs; high recall at a fraction of a brute-force scan's distance computations, with meas…

   Python

3. constrained-generation-library constrained-generation-library Public

   A structured-generation library built from scratch in pure Python: compiles a JSON Schema into a finite automaton and masks a model's tokens so output is schema-valid by construction (100% valid JS…

   Python

4. corpus-dedup-library corpus-dedup-library Public

   A corpus near-deduplication and quality-filtering library built from scratch in pure Python using MinHash and LSH; perfect precision and recall on a labeled corpus while comparing a tiny fraction o…

   Python

5. training-provenance training-provenance Public

   Content-addressed provenance tracking for machine learning pipelines: a deduplicating artifact store, a content-addressed lineage DAG, and bit-for-bit reproducibility and integrity verification.

   Python

6. transformer-from-scratch transformer-from-scratch Public

   A reverse-mode automatic differentiation engine and a decoder-only transformer language model, built from scratch in pure Python with no numerical libraries; gradients verified against finite diffe…

   Python