- Rating
- 2
- Confidence
- 4
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
This paper presents ADAPT, a text augmentation framework that trains an LLM generator using feedback from a downstream student model. It combines student prediction difficulty, semantic similarity, and self-validation in a group-relative policy optimization framework. The method is applied to three existing augmentation approaches and evaluated on five NLP tasks with different generator sizes and training set sizes. The authors report improvements in downstream performance, data diversity, and sample efficiency.
Critical strengths
-
The motivation is reasonable. Static prompting does not account for how the student changes during training, and using student feedback to update the generator is a natural way to address this limitation. ADAPT works with existing augmentation methods and does not require changes to the downstream model architecture.
-
The experiments cover a relatively broad range of settings, including five NLP tasks, three generator families, and four training set sizes. The authors also examine augmentation budgets, individual components, generated text quality, and computational costs. The reported gains for smaller generators are interesting, particularly in low-resource settings.
Critical weaknesses
-
The contribution of policy optimization is not clearly isolated. ADAPT combines student-guided selection, self-validation, and iterative pooling, but lacks a frozen-generator baseline using the same scoring and filtering. It remains unclear whether the gains come from policy learning or simply selecting better samples.
-
Several experimental results are inconsistent. Table 12's Small-regime results match Table 3's Medium-regime results rather than Table 2. Table 6 reports QA/NER performance drops after removing a reward whose weight is zero. Some reported standard errors, such as 99.1 ± 2.35 over ten runs, are also mathematically incompatible with bounded F1 scores.
-
The optimization procedure contains inconsistencies. Section 3 and Algorithm 1 handle unvalidated candidates differently during advantage clipping. The LoRA adapter is also reinitialized at each outer iteration, raising questions about how policy learning progresses across iterations.
4.The efficiency comparison is incomplete. ADAPT requires repeated student training, candidate generation, validation, and policy updates, but Table 5 only reports per-sample generation time. Without accounting for total computational cost, the claimed efficiency advantage remains uncertain.
Questions for authors
-
How much improvement does policy optimization provide beyond student-guided sample selection? Would a frozen generator with the same candidate pool size, difficulty scoring, and validation procedure achieve similar results?
-
Could the authors clarify the discrepancies among Tables 2, 3, 6, and 12? In particular, why does removing the similarity reward affect QA and NER when its weight is zero, and how were the standard errors calculated? The results from individual runs would help clarify these issues.
-
Which advantage clipping rule was used in the actual implementation? Section 3 and Algorithm 1 seem to handle unvalidated candidates differently. Also, since the adapter is reinitialized at each outer iteration, is the accumulated data pool the only information carried over from previous iterations?
-
Does ADAPT still outperform the baselines when the total computational cost is matched, including student retraining, candidate generation, policy optimization, and validation? The current comparison mainly controls the number of accepted samples rather than the overall cost.
-
How well does self-validation work when the generator is explicitly optimized to produce examples that fool the student? Since the same model generates and validates the samples, the two processes may share errors. Is there evidence that this does not lead to incorrect labels, especially for QA and NER?