Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
List of papers on hallucination detection in LLMs.
| Date | Stars |
|---|---|
| 2026-07-24 | 1119 |
| 2026-07-25 | 1119 |
| 2026-07-28 | 1119 |
| 2026-07-30 | 1119 |
| 2026-07-31 | 1118 |
| 2026-08-06 | 1118 |
Today
— stars today
This week
-1 stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
# awesome-hallucination-detection [](https://github.com/EdinburghNLP/awesome-hallucination-detection) [](https://opensource.org/licenses/Apache-2.0) [](https://github.com/EdinburghNLP/awesome-hallucination-detection/pulls) [](https://github.com/EdinburghNLP/awesome-hallucination-detection) [](https://github.com/EdinburghNLP/awesome-hallucination-detection) ## Papers and Summaries ### [Verifiable Rewards Beyond Math and Code: Lightweight Corpus-Grounded Process Supervision for Factual Question Answering](https://arxiv.org/abs/2605.29648) - **Metrics:** Exact Match (EM); training cost / wall-clock speedup (×) vs. neural-verifier baselines - **Datasets:** Five knowledge-intensive / factual QA benchmarks (incl. TriviaQA; 30 model×benchmark cells over six 3B–14B instruction-tuned models) - **Comments:** Proposes **CorVer** (Corpus Verify), a lightweight, plug-in *process reward* for RL fine-tuning (**GRPO**) of LLMs on factual QA that replaces neural verifiers (NLI, LLM-as-judge, retrieve-and-grade) with a **corpus-grounded** signal derived from **Wikipedia co-occurrence counts**. A small (0.5B) extractor turns each generated sentence into entity/claim spans whose corpus co-occurrence — queried via an **Infini-gram** index — yields sentence-level credit, which is mapped to **token-level advantages** and combined with response-level judge and format rewards in a GRPO update. Beats the raw baseline in every one of 30 cells (six instruction-tuned models, 3B–14B; five QA benchmarks), with an average **+4.1 EM on TriviaQA**, and outperforms four neural-verifier baselines in 18/20 cells at **4.8–8.4× lower training cost**. Code at https://github.com/shichengf/CorVer. (arXiv 2026) ### [REFUTE: Scientific Critique & Epistemic Calibration Benchmark](https://bgpt.pro/refute) - **Metrics:** Truth Score (composite), critique skill, calibration/Brier, forced-choice flaw accuracy, planted-flaw soundness accuracy, missing-evidence refusal rate - **Datasets:** REFUTE v3 (`BGPT-OFFICIAL/refute`; 320 scored questions from recent science paper summaries) - **Comments:** Open Apache-2.0 benchmark from BGPT for whether LLMs critique recent science summaries without quietly overclaiming. Core finding: scientific-sounding ≠ scientifically honest — critique skill and calibration dissociate. Current complete board tops out around 74.5 Truth Score. Judge-free Inspect / lm-eval adapters. Complements factuality/hallucination suites by testing overclaim correction and missing-evidence refusal on science summaries. ([Site](https://bgpt.pro/refute), [Dataset](https://huggingface.co/datasets/BGPT-OFFICIAL/refute), [Leaderboard](https://huggingface.co/spaces/BGPT-OFFICIAL/refute-leaderboard), [Preprint package](https://github.com/connerlambden/refute-inspect/releases/tag/v3.0.0-preprint)) (2026) ### [QuCo-RAG: Quantifying Uncertainty from the Pre-training Corpus for Dynamic Retrieval-Augmented Generation](https://arxiv.org/abs/2512.19134) - **Metrics:** Exact Match (EM), token-level F1 - **Datasets:** HotpotQA, 2WikiMultihopQA, ASQA (long-form), PubMedQA (biomedical) - **Comments:** Proposes **QuCo-RAG**, a dynamic RAG framework that determines *when* to retrieve by quantifying uncertainty from **objective pre-training-corpus statistics** rather than ill-calibrated model-internal signals (logits, entropy). Two-stage detection: (1) *Pre-Generation Knowledge Assessment* — queries entity frequencies in the pre-training corpus and triggers retrieval when entities are low-frequency (long-tail knowledge risk); (2) *Runtime Claim Verification* — extracts knowledge triplets from eac
Excerpt of 92,600 characters
Read on GitHub
Pasquale Minervini · @EdinburghNLP · United Kingdom
131
14
5
4
3
2
2
2
2
Allen Schmaltz
1
1
Jiaxin Zhang · United States
1
1
1
1
1
1
1
1
1
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:de5a4e6eaea2f2cd, readme:llm as judge, readme:leaderboard, name:hallucination detection
matched fp:de5a4e6eaea2f2cd, topic:nlp