GeneZip-636M-128K
GeneZip pretrained DNA language model, published under its Camera Ready name. This repository contains pretrained weights, not a downstream task checkpoint.
- Pretraining context: 128,000 bp.
- Base compression target: 32 bp/token (not the measured compression ratio).
- Source: jzshared/GeneZip-7R-LT-32BPT-128K-100B-650M.
- Weight SHA-256:
a70a35232d1bbcdc7c0a555e1fdcd279ffdf147a4029430e1874e9cddd4607e0. - Weight bytes are unchanged from this pinned source. The Hydra configuration is reduced to architecture, routing and scientific metadata.
Code and usage
Use the GeneZip Camera Ready code from AndyJZhao/GeneZip. The CR code update is being prepared separately; these weights have been checked with its local release candidate. This is a custom GeneZip checkpoint, not a Transformers AutoModel package.
from huggingface_hub import snapshot_download
checkpoint = snapshot_download("andyjzhao/GeneZip-636M-128K")
The release code accepts ckpt=andyjzhao/GeneZip-636M-128K for downstream training. For an explicit
local download, pass the returned directory as ckpt. The pretraining evaluator
uses pretrained_ckpt with the matching model/context configuration.
hydra_cfg.yaml records the exact architecture and routing bounds.
provenance.json records the immutable source revision and source weight hash.
The repository contains no downstream fine-tuned weights or optimizer state.
Intended use and limitations
Research on genomic sequence modeling and the GeneZip DNALongBench recipes. Short smoke checks establish loading and execution; they are not new measurements of full benchmark performance. Refer to the code's result mapping for adopted runs.