1
Show HN: Sketchlog – 100M events compressed to 93 KB using streaming sketches
流式草图技术将1亿事件压至93KB，极致压缩比令人惊叹。
https://github.com/SBALAVIGNESH123/sketchlog Comments URL: https://news.ycombinator.com/item?id=48507112 Points: 1 # Comments: 0
2
Show HN: NEP – Ethereum JSON-RPC transform that beats ZSTD by 12%
为以太坊数据量身打造的结构化编码层，比经典ZSTD压缩再提升12-21%，区块链节点同步更快更省带宽
Article URL: https://github.com/Louw115/nep-ethereum-compression Comments URL: https://news.ycombinator.com/item?id=48420973 Points: 3 # Comments: 1
3
Float8@2bits: Entropy Coding Enables Data-Free Model Compression
提出Float8@2bits方法，用熵编码在无数据场景下高效压缩模型，性能几乎无损。
arXiv:2601.22787v2 Announce Type: replace Abstract: Post-training compression is currently divided into two contrasting regimes. On the one hand, fast…
4
Zipping the Thought: When and How Compressed Reasoning Data Works in LLM Post-Training
探明压缩推理数据在LLM后训练中的生效条件与方式，为效率优化提供理论支撑。
arXiv:2605.28008v1 Announce Type: new Abstract: Large language models (LLMs) can now solve complex problems through long chain-of-thought (CoT) reason…