general-purpose-factuality-verifiers-entailment-consistency-ragGeneral-purpose factuality verifiers (entailment-, consistency-, and RAG-trained) and zero-shot frontier LLM judges perform near chance at spotting unsupported claims in scientific peer reviews that must be checked against the full submitted paper, because they cannot separate paper-unsupported assertions from legitimate evaluative critique; small models fine-tuned on in-domain examples beat them by a wide margin.
Ingested from a paper but not yet reviewed by a human. It is deliberately inert: it does not move any technique’s standing, does not count toward the backtest, and is excluded anywhere a claim would carry weight. Read the source before relying on it.
Capability: Stating false facts confidently
Observed on
Peer reviews of ML conference papers (ICLR/NeurIPS on OpenReview) with synthetically injected hallucinations; detection at review and sentence level with BM25 or full-paper evidenc.
Sources
- supportsHalluPeer: A Taxonomy-driven Benchmark for Detecting Hallucinations in Scientific Peer ReviewsMeasured on 12K papers and 38K reviews; four specialized verifiers gave review-level MCC of at most 0.03, prompted LLMs including GPT-5.2 stayed at 0.15-0.25 review-level MCC, while a fine-tuned Qwen2.5-3B reached 0.69 and Qwen3-32B 0.81. Hallucinations are LLM-injected rather than naturally occurring, so difficulty may reflect the injection distribution; a small authentic-review check (20 annotated cases) showed the fine-tuned detector recovered all of them at 22% false positive rate.
Disagreeing is the most useful thing you can do here. Both sides of every contested claim in this catalog were assembled by the same person, which is its weakest point.
Related claims
- Used zero-shot as step-level action critics in long-horizon tool-calling tasks, frontier models are over-pessimistic — flagging a large share of correct actions and pushing the actor into revision loops that lower success below no critic at all — whereas small models fine-tuned on action-level verification rationales flag far fewer valid actions and raise repeated-run reliability.Using the tools it is given · unreviewed
- In synthetic hallucination benchmarks where faithful items are human-written and hallucinated items are LLM-rewrites, a detector's binary score partly measures human-vs-machine style rather than factuality: feeding faithful LLM rewrites of the same truthful articles makes some fine-tuned encoders flag nearly all of them as hallucinated.Whether the measurement made the finding · unreviewed
- When training small instruction-tuned models with reinforcement learning plus factual supervision, routing each atomic fact's verification score only to the tokens that produced it, and down-weighting verifier judgements that do not change when their key evidence is removed, improves factuality benchmark scores over trajectory-level or reasoning- step-level factual rewards.Checking claims against evidence · unreviewed
- For small instruction-tuned models (3-4B) answering closed-form true/false questions under a stated user opinion, scoring each sampled response by how much more common its answer is than the group itself predicted, and using that score as the GRPO reward, reduces answer flips under user pressure and raises accuracy without any labels, roughly matching label-supervised desycophancy fine-tuning at higher compute cost.Telling the user what they want to hear · unreviewed
- When research-idea forecasts are scored by whether a later paper matches them under an LLM judge rubric, higher scores partly reflect broader, less specific ideas: the backbone with the higher hit rate also produced measurably more general forecasts and had more retrieved candidate papers pass the gate, and tightening the specificity threshold does not separate the two.Whether the measurement made the finding · unreviewed
Notes
Drafted from the paper by a model and filed unreviewed. Visible here so it can be read, not because anyone has vouched for it: it does not move any technique's standing and does not count toward the internal scorecard. Drafted confidence: medium. Falsifier as drafted: An off-the-shelf verifier or zero-shot frontier LLM reaching review-level MCC comparable to the in-domain fine-tuned models on this or a similar peer-review grounding benchmark. Proposed technique, not catalogued: Taxonomy-driven hallucination injection for review verification benchmarks.