- Rating
- 3
- Confidence
- 4
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
This paper proposes SHEAR, a method that edits hidden states during reasoning while keeping the model weights fixed. It uses a small probe to estimate current answer correctness and a direction extracted from training traces to guide editing. Experiments on three reasoning models report accuracy gains on math, science, and code tasks.
Critical strengths
- The motivation makes sense and the writing and experiment settings are generally solid.
- The paper connects hidden-state probing with online editing in a clear way. The probe provides feedback to guide the model during generation.
- The signal comparison controls both feature size and classifier choice. This provides useful evidence that hidden states contain stronger correctness signals than the tested token statistics.
- The evaluation covers three models and several tasks in different domains. Transfer to science and code without extra training, together with repeated sampling runs, provides useful support for the reported gains.
Critical weaknesses
- Wrong attempts are not always harmful. Trying a wrong answer, finding the mistake, and changing direction can help reasoning [1, 2]. Labels based on the current answer do not distinguish useful exploration from harmful mistakes, so a low probe score does not necessarily mean editing is needed.
- The benefit of hidden-state access is not fully isolated. The paper lacks a direct comparison with a text-based correctness judge using the same editing rule. Comparisons with token statistics do not establish whether the gains require hidden-state information or could come from any reliable correctness signal.
- It is unclear what behavior the editing direction changes. I think it is somehow tricky for the method. The score plots and direction ablations do not show how editing changes the actual reasoning text. Paired outputs with and without editing are needed to show whether it corrects reasoning, reduces repetition, or changes stopping behavior.
- The treatment of segments without an answer is unclear. The paper does not explain how such segments are labeled or how intermediate and multiple candidate answers are handled. If unfinished reasoning is labeled incorrect, the probe may confuse normal progress with reasoning errors.
- The control rule needs more complete ablations. The paper does not separately test the need for high-confidence edits and variance-based scaling, or fully study sensitivity to editing strengths and thresholds. The fixed-positive-edit baseline changes timing, strength, and direction together, so it does not isolate the benefit of adaptive control.
[1] Siyuan Wang, Yanchen Liu, and Xiang Ren. Segment-Level Attribution for Selective Learning of Long Reasoning Traces.ICLR, 2026.
[2] Zhuoshi Pan et al. LEMMA: Learning from Errors for MatheMatical Advancement in LLMs. ACL 2025.
Questions for authors
- Can you distinguish harmful mistakes from useful failed attempts and self-correction? Starting from the same reasoning prefix, does editing at a low probe score improve final success compared with letting the model continue without editing?
- Can you replace the hidden-state probe with a text-based verifier while keeping the editing direction and control rule fixed? Please compare both accuracy and computational cost.
- I would really love to see if you can prove that the method is solid. Can you show paired outputs from the same reasoning prefix, with and without editing, including both successful and harmful edits? What does the direction change in practice: calculations, exploration, repetition, or stopping?
- How do you label segments containing no answer, only an intermediate result, or several candidate answers? Do you exclude them, label them incorrect, or reuse an earlier answer?
- Can you test each part of the control rule separately and vary its strengths, thresholds, and window size? How many positive and negative edits occur per problem, and how does this depend on difficulty?
Response to Reviewer fjV2 (4)
Response to Weakness 4 and Question 4
Thank you for the question; the appendix indeed devotes only one sentence to this. The rule follows the actual behavior of
extract_segment_answerandtrain_probe.py: it tries explicit patterns such as\boxed{}and "answer is / = N" in turn, and if none matches, it takes the last number in the segment. Therefore, for a segment containing only an intermediate result, its candidate is its last number, and when there are several candidates, the last one is likewise taken. Empty is returned only if the segment contains no number at all. A step that returns empty is labeled 0 and kept in the training set, andskipped_no_answeris triggered only when the problem itself lacks a gold answer. The statistics of the three types over the 40,492 steps are as follows:To your concern that "the probe treats normal progress as an error", the data answer as follows: the probe scores of no-candidate steps and wrong-candidate steps almost completely coincide, with mean scores of 0.079 versus 0.077 and an AUC between the two of only 0.522, which shows that the probe reads whether the answer is correct, not whether an answer has appeared; moreover, among no-candidate steps only 27.1% trigger a push-away, and the rest stay neutral. We also retrained the probe with alternative labeling schemes:
The current scheme is no weaker than the two alternatives, and it is the simplest. In the final version we will write the rules above into Sec. 3.
Response to Weakness 5 and Question 5
Experiment a: ablating each part separately. 7B, single sampling run; each row changes only one factor. Vanilla and full SHEAR are taken from Table 3.
AMC23 has only 40 problems, so one problem equals 2.5 pt, and the standard deviation reported in Appendix G is 3.0 pt; we therefore do not use this column as standalone evidence. On MATH-500 and GSM8K, the standard deviation of the difference between two single runs is 2σ, and the 95% significance thresholds are about 1.9 pt and 1.1 pt: removing the low-confidence edit costs 3.0 / 5.3 pt, and removing variance amplification costs 2.6 / 1.9 pt, so both exceed the threshold on both benchmarks; removing the high-confidence edit costs 1.2 / 1.5 pt, which exceeds the threshold only on GSM8K and is insufficient to be judged on its own on MATH-500. The sum of the three losses on MATH-500 is 6.8 pt, larger than the end-to-end net gain of 4.4 pt, which indicates that the three parts overlap, so this is not an additive decomposition.