Tree-Ising Baseline - GSM8K: leaderboard
Tree-Ising Baseline - GSM8K: GSM8K exact-answer accuracy (%). Tree-Ising arXiv 2611.14210v1 Table 7: unmodified Qwen3.5-2B on all 1319 GSM8K test problems, greedy decoding, thinking mode, 8192-token cap. Edited checkpoints are not the unmodified model; Table 3 base numbers are copied from Takehi et al. and are not admitted as author-run observations.
Metric: GSM8K exact-answer accuracy (%). Source: arxiv.org. 1 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.5 2B | 21.6 | #4849 |
Interactive version: theaggregate.ai/benchmark?slug=tree-ising-baseline-gsm8k · How It Works · Data refreshed daily, snapshot 2026-11-12.