Published July 27, 2026
| Version v1
Software
Open
CAESAR + EDIR-Bench: reference implementation, benchmark, and raw results
Authors/Creators
Description
Reference implementation of the CAESAR neuro-symbolic multi-agent framework, the EDIR-Bench generator with independent Z3 oracle, external-validity experiment code (AutoGen head-to-head, NATURAL PLAN, solver-only baseline), and the raw result records (JSONL) underlying all reported tables.
Files
caesar-edir-zenodo-v1.0.zip
Additional details
Dates
- Accepted
-
2026-07-27