- Rating
- 2
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 1
- Soundness
- 2
Read review
Paper summary
The paper constructs geometric features of token-wise hidden-state trajectories across transformer depth, using competitor-based subspaces, orthogonal alignment, and module-attribution statistics. A supervised validator detects hallucinations, and its event scores guide rollback and single-block intervention. Evaluation covers five LLMs and twelve detection benchmarks, with refinement tested on nine benchmarks
Critical strengths
-
The paper presents a concrete construction of depthwise geometric features and connects supervised detection to targeted inference-time intervention without updating the base LLM.
-
Evaluation spans five LLMs and twelve benchmarks across several task categories, with feature-group ablations, refinement results, and measured extraction costs.
Critical weaknesses
-
Experiments. Broad coverage does not resolve the key attribution questions. ICR Probe (ACL 2025), which already studies cross-layer updates and module contributions, is omitted. Matched-supervision comparisons with ordered hidden states or simple differences, and ablations of subspace construction, alignment, and path integration, are needed; feature-group ablations cover only HaluEval QA. Refinement lacks matched regeneration-only and random/fixed-site controls, so improved outputs do not establish causal localization. Repeated-run uncertainty and human checks of judge labels are missing; a shared judge does not rule out differential bias. A newer open-weight model would also help test applicability.
-
Presentation. Even after careful reading, dense multi-index notation and abrupt transitions make the pipeline and its assumptions difficult to follow. The introduction overstates the static/dynamic distinction and moves too quickly from detection to error localization. Invariance claims need qualification for median centering, direction sampling, and weak-overlap identity resets. Online monitoring claims should be distinguished from replay-based evaluation, and the twofold decode bound from total runtime including extraction, fitting, and calibration. Figure 1(b–c) contains blurry elements, and the conclusion section opens with an editing fragment. A notation table and clearer separation of definitions, assumptions, and guarantees would help.
Questions for authors
-
How does the method compare with ICR Probe and similarly sized supervised detectors using ordered hidden states or simple differences? Which gains require subspaces, alignment, and path integration beyond HaluEval QA?
-
Does refinement outperform matched regeneration-only and random/fixed-block interventions? What supports interpreting the highest event score as an error source?
-
Please provide repeated-run uncertainty and manual label checks, and qualify the claim that judge noise cannot favor the method. Can applicability be checked on a newer open-weight model?
-
Please reconcile invariance with median centering, sampling, and identity resets, distinguish replay from online monitoring, and account for all overheads in the cost bound.
8 comments
Question 2
We thank the reviewer for asking for these controls, and we agree that a lower hallucination ratio after refinement does not by itself establish that the selected event is the source of the error. We have revised both the evidence and the claim.
Controls. We added the controls you requested to Table 2. On the same flagged responses, with the same rollback position and the same decoding budget, we compare the validator-selected block with regeneration without intervention (Regen-only), the same clamp on a random block (Random layer), and the same clamp on the middle block (Fixed layer). The table gives the mean hallucination ratio (%) over the eight refinement benchmarks.
| Base LLM | Before | Regen-only | Random layer | Fixed layer | Ours |
|---|---|---|---|---|---|
| Qwen2.5 | 23.01 | 23.92 | 24.04 | 21.62 | 17.86 |
| Gemma2 | 20.99 | 21.47 | 21.87 | 21.95 | 18.55 |
| Phi-3 | 19.86 | 23.92 | 22.66 | 21.35 | 14.05 |
| LLaMA3 | 26.27 | 28.54 | 29.01 | 28.62 | 22.71 |
| Mistral | 29.55 | 33.24 | 34.64 | 32.80 | 23.01 |
| All | 23.93 | 26.22 | 26.44 | 25.27 | 19.24 |
The validator-selected block gives the lowest ratio in 33 of 40 cells and never raises the ratio above its initial value, while the controls raise it in 24 (Regen-only), 29 (Random layer), and 25 (Fixed layer) of 40 cells. Since all variants share the rollback position, this comparison isolates the choice of block, and the gain comes from where we intervene rather than from resampling.
Interpretation of the top event. We do not interpret the highest-scoring event as the error source, and we have removed wording that suggested it. The validator is trained with response-level labels, so the top event marks where the evidence for hallucination peaks, which may come after the error first appeared. We now describe it as an intervention site, an effective place to correct the generation, and the controls above support this narrower claim. We also compared it with intervening at the first event above the decision threshold, on TriviaQA and HaluEval QA for all five models (Appendix E.1, Table 5). The earlier site gives the lower ratio in 3 of 10 cells, which shows that some errors do begin earlier, and the top event gives the lower ratio in the other 7.
Revisions. Table 2 adds the three controls. The Introduction (first and third paragraphs), the overview of the method section, and "Signature-guided single-block refinement" describe the selected event as an intervention site, and the Introduction and the refinement section state that it is not claimed to be the origin of the error. The term "culprit" is replaced. Appendix E.1 adds the earliest-site comparison.
Question 3
Repeated-run uncertainty. We thank the reviewer for these requests. We retrained our validator three times with different random seeds for every model and dataset combination, and Table 1 now reports the mean standard deviation. The standard deviation ranges from to AUROC across the 60 combinations, with a mean of , and is below in 38 of them. The detection results are therefore stable across training runs.
Manual label check. We manually labeled 100 outputs of Qwen2.5, 25 each from TriviaQA, HaluEval General, xSUM, and GSM8K, sampled to include both judge labels in near-equal numbers. The judge agrees with the human labels on 89 of 100 outputs, with Cohen's . Of the 11 disagreements, 7 are hallucinations the judge missed and 4 are false alarms.
| Judge: Halluc. | Judge: Not halluc. | |
|---|---|---|
| Human: Halluc. | 44 | 7 |
| Human: Not halluc. | 4 | 45 |
On judge bias. We agree and have qualified the claim. A shared judge cancels label noise that is unrelated to the detectors, but, as you note, it does not rule out bias that correlates with what a particular detector reads, and for refinement the judge scores different texts before and after the intervention. The revised text no longer states that judge noise cannot favor our method, and states explicitly that we do not claim immunity to judge error. What remains is a narrower observation about one asymmetry we can identify. The supervised detectors, ours included, train on the noisy labels that the score-based baselines only face at evaluation, so noise of this kind lowers rather than raises our advantage. The manual check above indicates the size of the noise, with 89 of 100 labels in agreement.
Revisions. Table 1 reports the mean standard deviation over three training runs for our validator. The manual check is added as Appendix E.3 with Table 7. The "Labels and baselines" paragraph in §4 is revised accordingly.
Question 1
We thank the reviewer for pointing us to ICR Probe, which is indeed closely related. We added it to Table 1 for all five base LLMs and twelve benchmarks. We use the official implementation and probe with the paper's settings, trained on the same training split as our validator. Our validator exceeds ICR Probe in 44 of 60 combinations, by +1.20 AUROC on average, and has the higher mean on every base LLM.
Matched-supervision comparison with ordered hidden states and simple differences. We replaced our signatures with two inputs read by the same GRU validator under the same training setup, so the detector, its size, and its supervision are matched. The first is the simple differences of consecutive hidden states (Simple diff.), and the second is the ordered hidden states themselves (Ordered states). Both fall well below our signatures in all 10 cells, by 14.5 and 14.3 AUROC on average.
Construction ablations beyond HaluEval QA. We extended Table 3 from HaluEval QA to HaluEval General, with the feature groups, the three construction steps, and the two replacements above. Each construction step lowers AUROC when removed. Fixing a single subspace for all depths (w/o Moving subspace) lowers it in 10 of 10 cells by 6.0 on average, replacing the path integral by the endpoint linearization (w/o Path integration) in 10 of 10 by 4.5, and removing the transport (w/o Transport) in 9 of 10 by 4.4. The one exception is Phi-3 on HaluEval General, at +0.6. The table gives the mean AUROC over the five base LLMs from a single training run.
Revisions. ICR Probe is added to Table 1 and to the baseline list in §4. Table 3 now covers HaluEval QA and HaluEval General and includes the construction steps and the two replacements, and §4.3 discusses them.