Published August 10, 2026
| Version v2.0-paper2
Software
Open
arqFranciscoLeon/evollm: v2.0 — Paper 2 replication package (Phase 2a)
Authors/Creators
Description
Replication package for "Not a Monolith: Lab-Level Divergence in the Cooperative Equilibria of Chinese Frontier LLM Agents".
Contains the 24 Chinese strategy libraries (DeepSeek V4 Pro, Qwen3-Max, Kimi K2.5, GLM-5.1), the 12 head-to-head tournament matrices, the n=500 Moran equilibria for all 48 conditions, the pre-registered converter robustness check of Section 4.7, and PHASE2_PREREG.md with the exact served model identifiers.
Supersedes v1.1-erratum, which covered Paper 1 only.
Files
arqFranciscoLeon/evollm-v2.0-paper2.zip
Files
(4.8 MB)
| Name | Size | Download all |
|---|---|---|
|
md5:83c639dd1e3075e39430b6d756534af8
|
4.8 MB | Preview Download |
Additional details
Related works
- Is derived from
- Preprint: arXiv:2501.16173 (arXiv)
- Is supplement to
- Software: https://github.com/arqFranciscoLeon/evollm/tree/v2.0-paper2 (URL)
- Preprint: arXiv:2605.29874 (arXiv)
Software
- Repository URL
- https://github.com/arqFranciscoLeon/evollm
- Programming language
- Python
- Development Status
- Active