- Rating
- 3(+1)
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
The paper presents achieves competitive results across multiple LVLMs and benchmarks. However, the methodological novelty is somewhat limited, and the relationship between the proposed geometric measurements and actual semantic grounding is not sufficiently established. Furthermore, inconsistencies in the detection targets across benchmarks make the broader generalization claims difficult to assess.
Critical strengths
- The paper providing a potentially useful perspective beyond conventional attention-based hallucination detection.
- The paper provides component ablations, layer-wise analyses, and visualizations that help characterize the complementary contributions of SAR and EMS.
Critical weaknesses
- The paper does not sufficiently demonstrate why these specific formulations are necessary compared with simpler combinations of attention, semantic similarity, and FFN-based features.
- The COCO experiments detect hallucinated object mentions, whereas the CLEVR and AMBER experiments classify whether generated answers match ground-truth answers. In the latter settings, the extracted target is the first answer token, making it unclear whether the method consistently measures object-specific grounding.
Questions for authors
- Can the authors compare SAM against simpler alternatives that directly concatenate attention, semantic-support, and FFN-alignment features under the same downstream classifier?
- How do the authors justify using first-answer-token correctness as a proxy for hallucination detection on CLEVR and AMBER? Would the results remain consistent when targeting the queried object or relation explicitly?
- Can the authors provide significance tests for the relatively small improvements over the strongest baselines?
5 comments
Response to Reviewer hV9S (2)
W1.2: On the Modeling Choices of SAR and EMS
demonstrate why these specific formulations are necessary
Response. Our core intuition goes beyond jointly using three types of information: it concerns two relationships—whether the visual sources contributing to attention writes support the current target, and how much the visual preferences of subsequent FFN updates deviate from this target-conditioned evidence. SAR and EMS translate these relationships into computable readouts. The theoretical analysis characterizes their properties, while the controlled comparisons in W1.1 & Q1 examine the roles of the individual modeling choices.
SAR: Preserving source-level attention–semantic correspondence. Separately aggregating WRITE magnitude and semantic support captures their overall strengths, but does not directly indicate whether strong writes originate from visual sources that support the target. For example, permuting semantic scores across visual sources leaves total semantic support unchanged, yet can alter the correspondence between semantic support and WRITE magnitude. SAR pairs the two before aggregation and therefore captures this distinction. As derived in Appendix A.2, Here, uses only total WRITE magnitude and mean semantic support, whereas the covariance term captures whether strong writes and strong semantic support occur at the same visual sources. The ablation therefore examines source-level pairing, while the variant retaining only examines the contribution of the original WRITE magnitude. Full SAR performs better in both comparisons, linking the observed benefits to specific modeling components.
EMS: Comparing target-conditioned evidence with update preferences, rather than merely measuring mean visual alignment. measures how similar the FFN update is to visual representations on average, without explicitly assessing whether those representations support the current target. For example, two updates can have the same mean visual alignment while favoring target-supporting regions and irrelevant background regions, respectively. Concatenating this mean with separately aggregated allows the MLP to learn statistical relationships among the aggregate features, but the omitted source-level correspondences cannot be determined from these aggregates alone. EMS instead retains the distribution of FFN preferences over visual sources and compares it with target-conditioned visual evidence, directly examining which visual sources the update favors and how these preferences deviate from the evidence supporting the current target. The geometric term further accounts for representation distances between visual sources, while the state term incorporates their compatibility with the current prediction state. The decomposition and geometric lower bound in the original manuscript characterize these properties. EMS improves detection performance on both models, supporting the value of this target-conditioned comparison over mean FFN alignment.
Default implementation: Cross-task performance and interpretability. SAM-V2 retains EMS and changes only the fusion of attention and semantic support. It therefore primarily examines the choice between SAR and the fusion of separately aggregated features. SAM-V2 performs better on COCO, but this advantage does not consistently hold across the evaluated VQA settings. As shown in W2 & Q2, SAM outperforms SAM-V2 in the All Answer, Pred-Yes, and Pred-No settings on Qwen3. On LLaVA, SAM performs better in Pred-Yes and Pred-No, while SAM-V2 performs better in All Answer.
The original visualizations further illustrate the meanings of the two readouts. The heatmaps show target-conditioned visual evidence and its spatial relationship with FFN preference. Figure 3 shows generally higher SAR for real objects and larger EMS for hallucinated objects across a broad range of layers. In contrast, the uncalibrated WRITE magnitude in the appendix exhibits substantial class overlap and local reversals. These observations align with our intuition: examining both whether visual writes receive target-semantic support and how subsequent updates deviate from that evidence.
We therefore retain SAR and EMS as our default: SAR explicitly preserves source-level attention–semantic correspondence, while EMS explicitly compares target-conditioned evidence with update preferences, and the corresponding controlled experiments support these modeling choices. We appreciate the reviewer's suggestion and will report these comparisons in full in the revised manuscript, explaining our default choice through cross-task performance and interpretability. We will also report SAM-V2, which retains EMS, as another effective feature-fusion implementation.
Response to Reviewer hV9S (3)
W2 & Q2.1: On Answer Correctness as a Proxy for Hallucination Detection
Why is the correctness of the first answer token used as a proxy for hallucination detection?
Response. We thank the Reviewer for this helpful question.
This choice follows prior work evaluating object hallucinations through existence questions. InsLen applies CLS and CCS to Yes/No tokens for object-existence hallucination detection. TokenGrounding assigns labels based on answer correctness, using the "Yes" token for ADS and the object term in the question for CGC.
For object-existence questions, Yes/No represents the model's judgment about whether the queried object exists, conditioned on the image and the full question. Thus, answer-targeted SAR evaluates the semantically calibrated visual evidence supporting this judgment, while EMS measures the mismatch between the corresponding evidence and FFN preference.
W2 & Q2.2: On Object-Targeted Evaluation
Do the results still hold when explicitly targeting the object or relation in the question?
Response. Following the Reviewer's suggestion, we further use the queried object as the EMS target while keeping SAR answer-targeted, to directly examine the role of object-related visual evidence.
Setup. To further connect the QA evaluation with our proposed object-level interpretation, we conduct additional experiments on CLEVR. We use the object term in the question as the EMS target to measure object-related evidence–update mismatch, while SAR uses the generated Yes/No answer as its target to evaluate the semantically calibrated visual evidence for that answer. For SAM-V2, the two separate features replacing SAR also use the answer token.(The differences from the original manuscript arise from randomness: the original three seeds varied only MLP initialization with a fixed image split, while the additional experiments vary both.)
We evaluate three settings:
- All answers: All samples, with labels determined by agreement between the generated and ground-truth answers.
- Pred-Yes: Among samples predicted as Yes, we distinguish correct Yes answers (GT = Yes) from unsupported Yes answers (GT = No).
- Pred-No: Among samples predicted as No, we distinguish correct No answers (GT = No) from incorrect No answers (GT = Yes).
Each entry in the following tables reports AUROC / AUPR.
LLaVA-1.5-7B
| Method | All answers | Pred-Yes | Pred-No |
|---|---|---|---|
| SVAR | 61.42 / 63.83 | 63.11 / 64.38 | 70.82 / 75.41 |
| DHCP | 56.30 / 58.35 | 61.86 / 62.86 | 63.92 / 67.68 |
| ProjectAway | 47.04 / 50.85 | 45.57 / 49.42 | 53.50 / 59.47 |
| TokenGrounding | 54.56 / 55.59 | 58.69 / 60.35 | 64.58 / 70.43 |
| InsLen | 46.37 / 50.20 | 45.19 / 49.11 | 52.78 / 59.00 |
| SAM | 65.36 / 67.88 | 65.39 / 67.81 | 71.38 / 75.62 |
| SAM-V2 | 66.12 / 68.73 | 64.12 / 67.24 | 70.70 / 73.68 |
Qwen3-VL-8B
| Method | All answers | Pred-Yes | Pred-No |
|---|---|---|---|
| SVAR | 66.49 / 98.17 | 68.03 / 98.44 | 75.64 / 98.44 |
| DHCP | 55.72 / 96.89 | 64.74 / 98.16 | 61.47 / 97.07 |
| ProjectAway | 44.28 / 95.96 | 46.39 / 97.26 | 44.26 / 94.80 |
| TokenGrounding | 69.79 / 98.31 | 62.07 / 98.39 | 82.78 / 98.94 |
| InsLen | 51.20 / 96.62 | 49.21 / 97.61 | 52.34 / 95.64 |
| SAM | 91.31 / 99.59 | 90.17 / 99.61 | 92.98 / 99.62 |
| SAM-V2 | 90.30 / 99.41 | 85.77 / 99.38 | 92.44 / 99.59 |
Observations. On LLaVA, SAM improves overall AUROC over SVAR by 3.94 percentage points and achieves better results in both answer subsets. Compared with SAM-V2, SAM improves AUROC by 1.27 and 0.68 points on Pred-Yes and Pred-No, while SAM-V2 performs better overall.
On Qwen3, SAM achieves the highest AUROC in all three settings, outperforming SAM-V2 by 1.01, 4.40, and 0.54 points.
These results directly support combining object-related evidence–update mismatch with answer-targeted visual evidence to detect unsupported Yes answers and incorrect No answers. Together with the source-level pairing ablation and interpretability analysis in W1 & Q1, these results further support our SAR/EMS design and show that the method maintains its detection performance when explicitly targeting the queried object.
Remarks. We again thank the Reviewer for the careful comments, which help us better clarify our method design, QA evaluation, and the statistical evidence for the performance differences.
Response to Reviewer hV9S (1)
We thank Reviewer hV9S for the careful review and constructive suggestions on feature design, VQA targets, and statistical reliability. We will include these results and clarifications in the revised manuscript.
W1.1 & Q1: On SAR/EMS Design and Simple Feature Concatenation
We have added simple concatenation and source-level pairing ablations to examine feature fusion, the role of EMS relative to mean FFN alignment, and the contributions of SAR's individual components.
Setup. We conduct experiments on COCO using LLaVA-1.5-7B and Qwen3-VL-8B. Let I=∑mIm, S=∑mβm, and Fmean=M1∑mcos(u,vm), where Fmean measures the mean alignment between the FFN update and visual reference representations. To examine source-level pairing, we replace SAR with I(1+sˉ), where sˉ=M1∑msm and sm=log(1+Cβm). We also evaluate a variant retaining only the pairing term B=∑mImsm to examine the role of the original WRITE magnitude in SAR=I+B.(The differences from the original manuscript arise from randomness: the original three seeds varied only MLP initialization with a fixed image split, while the additional experiments vary both.)
Observations. Simple concatenation performs well: the fully concatenated baseline is competitive with SAM and achieves higher AUROC on Qwen3, while SAM-V2 obtains the highest mean performance in the table on both models.
The controlled comparisons further show that replacing mean FFN alignment with EMS while keeping (I,S) fixed improves detection performance on both models. With EMS fixed, full SAR outperforms variants that remove either source-level pairing or the original WRITE magnitude. These trends support target-conditioned evidence–update comparison, source-level pairing, and the retention of the original WRITE magnitude, respectively. The rationale for these modeling choices and our choice of the default implementation are discussed further in W1.2.
Q3: On the Statistical Reliability of Gains over Strong Baselines
Response. We thank the Reviewer for this helpful suggestion. We have conducted a paired bootstrap analysis comparing SAM with the strong baseline SVAR.
Setup. We use three random seeds, each controlling both the image split and MLP initialization. For each bootstrap iteration, we resample images from the test set of each seed, using the same sampled images for both methods. We then compute the AUROC difference for each seed and average the differences across the three seeds. We repeat this process 3,000 times to obtain the 95% confidence interval (CI) of the mean difference.(The differences from the original manuscript arise from randomness: the original three seeds varied only MLP initialization with a fixed image split, while the additional experiments vary both.)
Observations. SAM improves mean AUROC by 1.12 and 1.27 percentage points on LLaVA and Qwen3, respectively. Both 95% confidence intervals are above zero, supporting the average performance gains over SVAR under this evaluation setting.