- Rating
- 2
- Confidence
- 3
- Informativeness
- 1
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
The paper proposes an inference-free framework for efficient supervised fine-tuning (SFT). This is done by dynamically choosing which tokens need to be trained on. The first component of the framework is MINT, which uses DPP to select a token subset. The second is Tempered Cross-Entropy (TCE), a modified loss function that ensures diversity by considering alternative tokens. The results show that this framework can outperform traditional SFT across multiple LLM families and datasets.
Critical strengths
The paper addresses fine-tuning efficiency, which is becoming increasingly relevant as datasets and models grow. The experiments also cover a variety of model families.
Critical weaknesses
My main concern with this paper is that it does not cite or discuss two very relevant works.
-
Nagaraj et al., "TRIM: Token-wise Attention-Derived Saliency for Data-Efficient Instruction Tuning." ICML 2026. This paper addresses the same issue similarly - it uses attention-derived signals from a forward pass to evaluate token-level importance. Hence, it is very important for the paper to compare MINT to TRIM: how do they differ, how they are similar, a comparison of results if possible, and where each application lies.
-
Jia, Yaning, et al. "Which Tokens Should SFT Actually Learn? A Token-Trimming Perspective on Mathematical Reasoning." 2026. This paper also similarly scales SFT loss using the per-token logit gap, trimming both extremes of model confidence. How is TCE different from this paper? Can there be a comparison?
Questions for authors
Please provide a comparison with the papers mentioned in critical weaknesses and any other relevant papers.
6 comments
Response to Reviewer p3qy
Q1: Please provide a comparison with the papers mentioned in critical weaknesses and any other relevant papers.
We discuss the relevant work in the Related Work section and report the experimental results on LLaMA-3.1-8B below. Scores are unweighted averages over the eight downstream benchmarks; ssToken results are taken from Table 1 in our response to Reviewer #2, and Full Token + TCE and TrimSFT results are taken from Table 1 in this response.
Table 3. Average benchmark performance with CE and TCE across selection strategies.
| Selection strategy | CE | TCE | Gain from TCE |
|---|---|---|---|
| Full Token | 54.80 | 56.92 | +2.12 |
| Random | 54.50 | 56.26 | +1.76 |
| RHO-1 | 54.86 | 56.71 | +1.85 |
| TokenCleaning | 55.69 | 57.11 | +1.42 |
| DFT | 54.90 | 56.74 | +1.84 |
| EFT | 54.41 | 56.58 | +2.17 |
| ssToken | 55.00 | 56.41 | +1.41 |
| MINT | 56.46 | 58.47 | +2.01 |
Table 4. Average benchmark performance with different training objectives.
| Training objective | Average | Change from SFT |
|---|---|---|
| SFT (CE) | 54.80 | — |
| TrimSFT | 50.93 | -3.87 |
| ToFu | 53.17 | -1.63 |
| KL | 55.09 | +0.29 |
| GEM | 55.62 | +0.82 |
| DFT | 54.90 | +0.10 |
| NEFT | 53.97 | -0.83 |
| TCE | 56.92 | +2.12 |
We have expanded the ssToken baseline comparison in Section 4.2 (Table 1) and the selection-strategy and training-objective comparisons in Section 4.3 (Figure 6) and Appendix D.1 (Figure 10).
W1: Nagaraj et al., "TRIM: Token-wise Attention-Derived Saliency for Data-Efficient Instruction Tuning." ICML 2026. This paper addresses the same issue similarly - it uses attention-derived signals from a forward pass to evaluate token-level importance. Hence, it is very important for the paper to compare MINT to TRIM: how do they differ, how they are similar, a comparison of results if possible, and where each application lies.
We thank the reviewer for pointing out the related work TRIM. Although both methods leverage attention-derived signals, they address fundamentally different data selection problems: TRIM selects training samples, whereas MINT selects response tokens within fixed training samples. Therefore, a direct quantitative comparison is not well-defined due to their different selection units and budget definitions. Nevertheless, we clarify their similarities and differences from three perspectives.
(1) Problem formulation. TRIM evaluates individual token relevance and aggregates scores for sample selection. In contrast, MINT reformulates token selection as an instruction-aware subset utility maximization problem, explicitly modeling token interactions to select informative and complementary tokens under a limited budget. Thus, MINT's contribution lies not merely in exploiting attention signals, but in this new problem formulation.
(2) Task dependency. TRIM primarily targets task-specific selection using target-set examples to construct token fingerprints (with an in-domain variant using an internally sampled target pool). MINT is target-task-agnostic and requires neither external validation data nor task-specific fingerprints.
(3) Selection efficiency. TRIM requires separate LoRA warmup and forward-scoring stages. MINT directly reuses logits and attention signals from the standard SFT forward pass, requiring no separate warmup, external reference model, or additional inference passes.
We have added a discussion distinguishing TRIM’s sample-level selection from TokenFlow’s token-level selection in Appendix B (Related Work); please refer to this addition in the revised manuscript.
Thank you for providing these comparisons. Adding these to the paper will strengthen it. I will revise my score accordingly.
Response to Reviewer p3qy
Thank you for highlighting this related work. We clarify how TCE differs from TrimSFT and provide an additional experimental comparison below.
Part 1. Different optimization mechanisms
TrimSFT follows a token-level loss reweighting paradigm: it uses logit gaps to adjust the contribution of individual tokens while retaining the original ground-truth supervision. Thus, it only changes how strongly each ground-truth token is learned, without explicitly preserving plausible alternative continuations in the supervision signal. In contrast, TokenFlow introduces Tempered Cross-Entropy (TCE), which incorporates plausible alternatives from the model's predictive distribution alongside the ground-truth token. By enriching the supervision signal rather than merely reweighting token losses, TCE avoids unnecessarily suppressing semantically valid alternatives, providing an explicit mechanism for preserving response diversity.
Part 2. Additional Experiment
We compare Full-Token SFT with TCE against Full-Token SFT with TrimSFT, with benchmark and response-diversity results reported in Tables 1 and 2, respectively. TCE consistently achieves better average task performance across all three backbones. It also exhibits substantially higher response diversity on both AlpacaEval and GSM8K, as reflected by lexical and semantic diversity metrics. Moreover, its superior GSM8K pass@k performance demonstrates better coverage of correct solutions under repeated sampling. Finally, incorporating MINT further improves benchmark performance and GSM8K pass@k, highlighting the complementary contribution of diversity-aware token selection.
Table 1. Benchmark performance
Table 2. Response diversity and repeated-sampling performance
(a) Response diversity on AlpacaEval
(b) Response diversity on GSM8K
(c) GSM8K pass@k
We have added the TrimSFT comparison to Section 4.3 (Figure 6(c)) and Appendix D.1 (Figure 10(c)).