waldiez/slm-callsite-eval

NeurIPS 2026 SLM-Agents 工作坊论文的基准仓库 waldiez/slm-callsite-eval 在 README 中添加了 arXiv、Hugging Face 论文页面与数据集链接，以及 `load_dataset` 示例代码。

GitHub 项目说明: Benchmark, harness and scored records for "Not Every Call Needs a Frontier Model" (SLM-Agents @ NeurIPS 2026): per-call-site evaluation of 9 local and hosted LLMs across 5 call sites of the Wactorz home-automation agent, with live end-to-end validation.

星标

★ 3

复刻

0

语言

Python

许可证

Apache-2.0