InfiCoEvalChain: A Blockchain-Based Decentralized Framework for Collaborative LLM Evaluation
Yifan Yang, Jinjia Li, Kunxi Li, Puhao Zheng, Yuanyi Wang, Zheyan Qu, Yang Yu, Jianmin Wu, Ming Li, Hongxia Yang · 2026
Read InfiCoEvalChain: A Blockchain-Based Decentralized Framework for Collaborative LLM Evaluation by Yifan Yang, Jinjia Li, Kunxi Li, Puhao Zheng, Yuanyi Wang, Zheyan Qu, Yang Yu, Jianmin Wu, Ming Li, Hongxia Yang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-08229.txt.
Description
Rankings for language models look precise, but rerun runs tell a shakier story: the standard spread across ten runs of one model on HumanEval, 1.67 points, exceeds the 0.91-point gap among the top ten models. Yifan Yang and nine co-authors propose InfiCoEvalChain, a blockchain-based spread out method for joint model testing.
Free checkers across mixed hardware run large-scale benchmarks under a reward system that deters crooked taking part, turning testing from a central black box into joint checks. The design embraces hardware and parameter diversity instead of pretending it away. Testing teams and leaderboard users gain a method aimed at the uncomfortable fact that current rankings may be statistically precarious.
Catalog record
| CREATOR | Yifan Yang, Jinjia Li, Kunxi Li, Puhao Zheng, Yuanyi Wang, Zheyan Qu, Yang Yu, Jianmin Wu, Ming Li, Hongxia Yang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 50 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-08229