[EMNLP'25] s3 - ⚡ Efficient & Effective Search Agent Training via RL for RAG (RLVR for Search with Minimal Data)
-
Updated
Nov 5, 2025 - Python
[EMNLP'25] s3 - ⚡ Efficient & Effective Search Agent Training via RL for RAG (RLVR for Search with Minimal Data)
SoTo is a framework for test-time search in autoregressive (AR) image generation, decoupling AR priors, search algorithms, and verifiers for flexible composition via simple YAML configs.
Verifiable Credentials and Decentralized Identity Features Offered by EveryCRED
Simple images from CSV downloader that runs and records checksums on downloaded image folder.
Loop Architect — a verifier-first skill for designing safe AI agent loops, or refusing when a task shouldn't be looped. Works with Claude Code, Codex, and any skill-capable harness.
Learner-Verifier Framework for formal synthesis of Control Barrier Functions (CBF).
Composable, offline, stdlib-only verifier for re-derivable, attested artifacts — recompute the answer, don't trust the claim. Apache-2.0.
Open skills, harnesses, hooks, and verifiers for giving probabilistic AI a control layer.
eBPF verifier bypass PoCs on Linux LTS 6.8 - kernel privilege escalation via logic bugs & ISO-IEC TS 17961 violations.
Generate and save checksums for all (or certain) contents of given directory.
Verifiable environments for evaluating optimization & planning agents under real-world constraints (RL / RLVR)
DualityCert: verifier-gated LLM repair of Seiberg-duality claims in QFT (arXiv:2607.23614). First instance of the QFTCert program.
Draft verifier for machine-checkable elements of Saudi SCC transfer documentation.
Memory efficient mbox-type mailbox files comparator for data integrity verification. It compares full content of emails (headers and body) after canonicalisation, and optionally tolerates differences in the lengths of blank-line runs. Optionally, it can handle mboxes with differing message orderings while still providing efficient resource usage.
Resample or reroute after a weak-verifier stop? Pre-registered measurements of recoverable stopping debt on MBPP+, a two-sided action-support gate on BigCodeBench that fails closed, a LiveCodeBench observability ladder, and the exchangeable-actions reference showing realized-maximum gaps carry no selector signal. Artifacts for arXiv:2607.08665v3.
An AI agent harness for teaching — plan mode, coach verifier, and a mistakes-become-rules ratchet
To associate your repository with the verifier topic, visit your repo's landing page and select "manage topics."