- Rating
- 3
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
iSDFT moderates a demonstration-conditioned teacher through an information-constrained projection toward the current student, while separately anchoring the student to its frozen base model. Across four backbones and two specialisation tasks, it improves over SDFT in seven of eight settings, matches the remaining setting, and reports better aggregate capability retention.
Critical strengths
- The motivation distinguishes local teacher influence from cumulative drift. The projection has a clear mathematical interpretation, and separating its role from base anchoring provides a coherent approach to the stability–plasticity trade-off.
- The evaluation covers diverse backbones and capability benchmarks. Tables 11–12 provide useful component evidence: scheduled targets improve acquisition with or without anchoring, while anchoring generally improves retention. The paper also acknowledges limitations in sequential forgetting and reasoning evaluation.
Critical weaknesses
- Attribution to the information constraint. The ablations establish benefits from the scheduled target and anchor, but do not isolate tokenwise information control from simpler teacher warmup. Fixed information budgets still use the same adaptive projection, and Table 12 omits retention results for SDFT+anchor. Thus, the necessity of the local constraint beyond scheduling and anchoring remains uncertain.
- Interpretation of reasoning gains. Performance under a common 4096-token budget has practical value. However, Appendix D.6 documents substantial truncation, while the 32768-token follow-up reports lengths without correctness or an SDFT comparator. Improvements therefore do not yet distinguish retained reasoning ability from changes in response length or answer presentation.
- Reliability of modest differences. Table 2 specifies one training seed. Several acquisition and ablation improvements are small, and sign tests across related benchmark evaluations do not establish stability across training runs. The cross-model consistency is encouraging, but uncertainty in these differences remains insufficiently quantified.
Questions for authors
- In one representative setting, how does iSDFT compare with a simple time-scheduled geometric target under the same teacher, loss, and anchor? Reporting retention for the existing SDFT+anchor configuration would further clarify the projection's contribution.
- Could one representative mathematical evaluation compare Base/SDFT/iSDFT correctness at a longer budget with verified answer extraction? Otherwise, please consistently delimit the claims to performance under the evaluated generation budget.
- Could a small paired repeat of one representative SDFT/iSDFT setting quantify training variability? Existing repetitions would suffice; repeating the entire evaluation matrix is unnecessary.
3 comments
Interpretation of reasoning gains. Performance under a common 4096-token budget has practical value. However, Appendix D.6 documents substantial truncation, while the 32768-token follow-up reports lengths without correctness or an SDFT comparator. .... Could one representative mathematical evaluation compare Base/SDFT/iSDFT correctness at a longer budget with verified answer extraction? Otherwise, please consistently delimit the claims to performance under the evaluated generation budget.
We ran this evaluation on MATH-500, which showed the largest improvement among the additional mathematics and coding benchmarks: Qwen3-1.7B before and after Tool Use specialisation with SDFT and iSDFT, at 32,768 tokens instead of 4,096.
| Model | 4,096 tokens | 32,768 tokens |
|---|---|---|
| Base | 21.6 | 27.2 |
| SDFT | 59.0 | 59.8 |
| iSDFT | 60.6 | 61.2 |
With the increased budget, the base model achieves 27.2, 5.6 points higher than under the original budget, while SDFT achieves 59.8 and iSDFT achieves 61.2, outperforming the base model by 34.0 points and SDFT by 1.4. These results show that the improvements on MATH500 persist under a larger generation budget, suggesting that the gains are not just due to response truncation.
Reliability of modest differences. Table 2 specifies one training seed. Several acquisition and ablation improvements are small, and sign tests across related benchmark evaluations do not establish stability across training runs.
Could a small paired repeat of one representative SDFT/iSDFT setting quantify training variability? Existing repetitions would suffice; repeating the entire evaluation matrix is unnecessary.
We ran additional experiments across two backbone models, Qwen3-1.7B and OLMo3-7B-Think, and two tasks (Tool Use and Science) using three seeds 0, 1 and 42 under identical configurations for SDFT and iSDFT. The two methods share each seed, so the comparison is paired. iSDFT achieves higher mean accuracy in all four model–task settings. Across all configurations, the standard deviations are at most 1.3 percentage points. These results provide additional evidence of robustness to training stochasticity and show that the improvements of iSDFT are consistent across independent training runs.
Mean ± standard deviation over the three seeds:
| Backbone | Task | SDFT | iSDFT |
|---|---|---|---|
| Qwen3-1.7B | Tool Use | 52.6 ± 1.3 | 52.9 ± 1.1 |
| Qwen3-1.7B | Science | 54.4 ± 0.8 | 56.5 ± 0.7 |
| OLMo3-7B-Think | Tool Use | 51.5 ± 1.2 | 54.0 ± 0.9 |
| OLMo3-7B-Think | Science | 14.4 ± 1.0 | 17.9 ± 0.6 |
The new experiments isolate the contribution of each component. With the same schedule and the same total teacher information, sharing one coefficient across tokens loses most of iSDFT's Science gain over SDFT + anchor, so most of that gain comes from the per-token information constraint rather than the warm-up.
And with the anchor held fixed, the projection adds this acquisition without a net loss in retention, and the anchor provides the retention. The MATH-500 gains persist at 32,768 tokens, and the seed repeats show that the comparison with SDFT is robust to training stochasticity. We hope these results address the reviewer's questions.
Thank you for the additional experiments. The matched controls, longer-budget evaluation, and seed repeats substantially address my concerns. The retention results also clarify that the projection improves acquisition, while anchoring supports retention. I will maintain my rating of 3 in support of acceptance.
We thank the reviewer for the constructive review. Each question asked for a specific experiment, and we have run all; the results are reported below.
We added this comparison on Qwen3-1.7B. It keeps iSDFT as is, with the same teacher, loss, anchor and hyperparameters, but uses one coefficient for all tokens instead of the per-token λt: fixed λ = 0.5, fixed λ = 0.75, a linear warm-up λ: 0.25 → 1 over the first 150 steps (the same timing as our budget schedule), and a pooled budget. The pooled budget keeps iSDFT's ρ schedule and solves one λ per response, shared by all its tokens, so that each response receives the same total teacher information as under iSDFT. Only the distribution across tokens differs.
On Science, no single coefficient matches iSDFT. The pooled budget isolates tokenwise information control: it has the same budget schedule and the same total teacher information as iSDFT, and differs only in sharing λ across the tokens of a response. It is the strongest single-coefficient target, yet it trails iSDFT by 2.0 points on Science. The linear warm-up, the simple time-scheduled target, trails by 3.8 points, and the fixed coefficients by more. Each row is a single run, and Tool Use differences are at most four of 97 questions.
We evaluated SDFT + anchor on the same 32 retention evaluations as Table 12 (16 benchmarks after each of the two tasks; retention table below):
So the projection's contribution is better acquisition at equal retention, not better retention on its own, while the anchor helps with retention.
Retention table: Qwen3-1.7B with the anchor, under the full-teacher target (SDFT + anchor) and the information-proximal target (iSDFT); bold marks the better configuration for each task (Tool Use or Science).