- Rating
- 2
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
This paper proposes Risk-Aware Dynamic Action Allocation (RADA) for collaboration between a small language model and a large language model. Given an SLM-generated reasoning trajectory, RADA predicts both response-level and step-level risks and selects among three actions: accepting the SLM response, locally revising the trajectory from an identified risky step, or restarting with full LLM regeneration. It also introduces a stateful threshold that is updated across queries using calibrated average qualities of the selected actions.
The method is evaluated on mathematical and broader reasoning benchmarks with Gemma and Qwen model pairs. RADA generally achieves the highest average accuracy among the evaluated collaborative methods, with competitive estimated FLOPs. Ablations study global and local risk signals, intervention modes, risk-loss weights, threshold adaptation, and the local-risk window.
Critical strengths
-
The paper studies an important routing decision beyond binary escalation. Selecting among acceptance, local repair, and full regeneration is practically relevant because different reasoning failures require different intervention scopes. The comparison in Table 5 suggests that neither fixed intervention mode is sufficient.
-
The method combines global and local evidence in an interpretable framework. The joint risk estimator, structural step roles, and explicit routing conditions make the intervention policy easier to inspect than an entirely opaque end-to-end router. The evaluation covers multiple task groups and model pairs.
Critical weaknesses
- The dynamic threshold is not updated using observed response quality and is inherently order-dependent. At inference, the update uses a fixed calibration-set average for the selected action rather than the actual quality of the current output. Therefore, the state depends mainly on the sequence of selected actions, rather than realized historical performance. Moreover,
$
$
accumulates the quality difference in and then repeatedly accumulates into the threshold. This can make predictions sensitive to query order and clipping. Since evaluation processes examples sequentially in their original order, order robustness is essential to the claimed adaptive behavior.
- The empirical evidence does not quantify uncertainty and may involve test-set-guided configuration choices. Experiments use stochastic decoding with temperature , but report only point estimates from an apparently single risk-model training run. Some improvements, such as the -point average gain over COREA on broader reasoning with Qwen, may be within run-to-run variation. In addition, the paper uses MATH500 results to motivate the default loss weights and local-risk window, which raises a concern about test-set model selection unless these settings were fixed beforehand. Multiple seeds, paired confidence intervals, and a clear separation between validation and final testing are needed.
Questions for authors
-
Clarification: Why is the threshold update described as using historical quality feedback when is a fixed calibration-set average and no current-query outcome is observed? Are and reset between benchmarks or sessions?
-
Assessment-changing analysis: How sensitive are the results to test-query ordering? Please report results across several random permutations, including threshold trajectories, clipping frequency, action proportions, accuracy, and FLOPs.
-
Clarification and existing evidence: Were the loss weights and selected before evaluating MATH500? Please provide multi-seed or repeated-decoding results with paired confidence intervals for the main comparisons, especially the smaller gains on broader reasoning tasks.
3 comments
Response to Reviewer YPwK (Part 2)
Q3: Were the loss weights and fixed before MATH500 evaluation, and what repeated-run and paired-confidence evidence is available?
- Paired confidence intervals. We further completed instance-level paired comparisons between RADA and STEER on ACPBench, covering all 900 test instances across five random seeds (101, 202, 303, 404, and 505). For each seed, we categorized the paired predictions into four outcomes: only RADA correct, only STEER correct, both correct, and both incorrect. Based on these paired outcomes, we calculated the accuracy difference between RADA and STEER and its corresponding 95% confidence interval. The results are reported below:
| Seed | Total | Only RADA Correct | Only STEER Correct | Both Correct | Both Incorrect | 95% CI (RADA − STEER, pp) |
|---|---|---|---|---|---|---|
| 101 | 900 | 25 | 29 | 507 | 339 | [-2.05, +1.16] |
| 202 | 900 | 30 | 31 | 498 | 341 | [-1.81, +1.59] |
| 303 | 900 | 29 | 32 | 500 | 339 | [-2.04, +1.37] |
| 404 | 900 | 28 | 32 | 502 | 338 | [-2.13, +1.24] |
| 505 | 900 | 26 | 32 | 507 | 335 | [-2.33, +0.99] |
All five 95% confidence intervals include zero, indicating that the paired accuracy differences between RADA and STEER are not statistically significant at the 0.05 level on ACPBench. This analysis provides additional instance-level statistical evidence for a more comprehensive assessment of the performance differences between the two methods.
We note that, due to the limited time available during the rebuttal period, we were only able to complete the additional paired confidence interval analysis on ACPBench. On this benchmark, RADA's accuracy is slightly lower than STEER's, although the observed difference is small and not statistically significant. However, this result should be considered alongside the broader experimental evidence reported in the manuscript. RADA achieves higher average accuracy than STEER on both mathematical reasoning benchmarks (42.70% vs. 38.37%) and broader reasoning benchmarks (43.28% vs. 42.16%) with the Gemma 4B–12B model pair. These results demonstrate that, despite a marginal difference on ACPBench, RADA maintains stronger overall accuracy across the evaluated benchmarks.
Additional Clarifications, if Needed
We sincerely thank all reviewers again for their valuable feedback. We have provided point-by-point responses to the concerns raised in the reviews, together with additional clarifications and experimental results. As the discussion period is approaching its conclusion, we would like to ask whether there are any remaining questions or concerns that have not been fully addressed by our responses. Please feel free to let us know if further clarification is needed, and we will do our best to address them before the discussion period ends. We greatly appreciate your time and constructive feedback.
Response to Reviewer YPwK
We thank the reviewer for the constructive questions. We clarify the implementation details and present additional experimental evidence below.
Q1: What does “historical quality feedback” mean, and when are the gating states reset?
Q2: How sensitive are the results to query order, including threshold trajectories, clipping, action proportions, accuracy, and FLOPs?
Q3: Were the loss weights and w=5 fixed before MATH500 evaluation, and what repeated-run and paired-confidence evidence is available?
Together with the five-seed order-only results in Q2, this supports stability under the tested randomizations; order-only runs should not be conflated with independent decoding runs.