What it is
Hash-pinned reproduction bundle for a CASMI26 mass-spectrometry pipeline, packaging the exact public candidate pools, pretrained models, code, and manifests used in one Kaggle workflow. 134 hash-pinned files across 5 source folders: a 710,701-structure train+COCONUT pool, 105.9M mass-sorted PubChem structures, FPNet/ranker assets, v4b engine files, and optional ICEBERG reranking resources.
Key results
- 134 source files totaling 11,060,095,811 bytes
- 710,701 train+COCONUT candidate structures
- 105.9 million mass-sorted PubChem structures
Useful for reproducing and auditing a CASMI26 molecular identification pipeline from spectra to ranked structures.