- Rating
- 3
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 3(+1)
Read review
Paper summary
The paper proposes a training-free visual token pruning method for efficient VLM reasoning. It first constructs a cached pool of diverse visual tokens, then dynamically selects visible tokens from the pool based on current and historical attention during decoding. It also uses answer-consistency probes to terminate reasoning early when a stable answer is detected. Experiments show that it outperforms other pruning methods with shorter generation length on seven multimodal benchmarks.
Critical strengths
- The paper addresses an important practical problem in improving the inference efficiency of reasoning-capable VLMs. It highlights that computation saved by visual token pruning during prefill may be offset by increased decoding costs.
- The empirical evaluation is extensive, covering multiple VLM backbones, relevant pruning baselines, and different visual token budgets. The paper also reports generation length, inference latency, and TFLOPs, with component-level ablations.
Critical weaknesses
- The main comparisons don’t fully match the visual-token budgets. VISTA exposes c tokens per decoding step but caches a larger 2c reserve during prefill. Although the reported end-to-end speedups are encouraging, accuracy-latency comparisons across budgets could better establish the efficiency trade-off.
Questions for authors
- How frequently are 2c reserve tokens actually activated during decoding? Can these events be related to error recovery?
- Could the authors explain the substantial performance degradation compared to Vanilla (e.g., InfoVQA on Qwen3-VL-4B-Thinking and ChartQA on InternVL3.5-4B-Thinking) with representative failures?
4 comments
Rebuttal by Authors (2/2)
W1: Although the reported end-to-end speedups are encouraging, accuracy-latency comparisons across budgets could better establish the efficiency trade-off.
We appreciate your suggestion to provide accuracy-latency comparisons across budgets. We summarize the results across all seven benchmarks in Table R2. We observe that for VLMs with stronger reasoning capabilities, under aggressive visual-token pruning, the overall computational bottleneck shifts from prefill to decoding. The additional decoding overhead can outweigh the computational savings from prefill, leading to significantly higher total computational costs and latency, which remains consistent across different budgets. Furthermore, under same prefill computational cost (2.7 TFLOPs), VISTA at a 10% decoding visual-token budget still achieves higher accuracy than the best baseline at 20% (65.6% vs. 62.6%). Meanwhile, VISTA reduces the average generation length by 39.1% and end-to-end latency by 38.2%. Thus, suggesting that VISTA achieves a more favorable accuracy-latency trade-off by effectively mitigating reasoning inflation while maintaining high accuracy.
Table R2. Accuracy-latency comparisons across budgets. Average over all seven benchmarks
| Method | Budget | Acc (%) | Length | Prefill TFLOPs | Decoding TFLOPs | Total TFLOPs | Time (s/q) |
|---|---|---|---|---|---|---|---|
| DivPrune | 10% | 52.3 | 2,230 | 1.6 | 24.0 | 25.6 | 30.4 |
| ZOO-Prune | 10% | 50.0 | 2,543 | 1.6 | 27.6 | 29.2 | 37.2 |
| MMTok | 10% | 48.5 | 2,261 | 1.6 | 24.2 | 25.8 | 32.5 |
| VISTA | 10% | 65.6 | 1,205 | 2.7 | 12.6 | 15.3 | 15.2 |
| DivPrune | 20% | 62.6 | 1,979 | 2.7 | 21.6 | 24.3 | 24.6 |
| ZOO-Prune | 20% | 62.3 | 2,077 | 2.7 | 22.5 | 25.2 | 26.6 |
| MMTok | 20% | 59.6 | 2,089 | 2.7 | 22.8 | 25.5 | 26.9 |
| VISTA | 20% | 71.3 | 1,073 | 5.1 | 11.3 | 16.4 | 13.4 |
Thanks for the detailed response. My major concerns are solved and I am happy to keep my acceptance recommendation.
Many thanks for your positive feedback and for the time and effort you have devoted to improving our work. We sincerely appreciate your constructive comments and valuable suggestions.
Rebuttal by Authors (1/2)
We sincerely appreciate your acknowledgment and constructive comments to improve our work. Below, we will address your concerns point by point. All additional results and analyses have been incorporated into the revised manuscript and highlighted in orange.
We appreciate your comment on reserve-token activation and error recovery. Here,
Table R1reports the rate of decoding steps with at least one activation and initially invisible reserve tokens activated at least once during decoding on Qwen3-VL-4B-Thinking under a 10% visual-token budget. We define an activation as a reserve token transitioning from the invisible state to the visible set. On average, activations occur in 85.7% of decoding steps, and 92.4% of initially hidden reserve tokens are activated at least once, indicating a continuous revisit of reserve tokens throughout decoding rather than remaining as an unused cache.We further investigate the relationship between dynamic activation and error recovery using the Fixed variant, which does not activate invisible reserve tokens during decoding, under otherwise identical settings (consistent with
Table 7in revised manuscript). Among questions answered incorrectly by Fixed, an average of 28.1% are corrected by VISTA, which improves average accuracy from 55.0% to 61.9%. These results suggest that dynamic activation effectively recovers critical visual evidence and thereby recover errors. Following your suggestion, we have incorporated these results and analyses into the revised manuscript (Lines 501andTable 8).Table R1. Reserve-token activation and error recovery on Qwen3-VL-4B-Thinking with a 10% decoding visual-token budget. Fixed disables dynamic activation. Wrong → Right measures corrected answers across all questions, while Fixed errors recovered measures the recovery rate among Fixed's errors.
We appreciate your suggestion on representative failure cases. For InfoVQA and ChartQA, questions typically require identifying specific labels in small regions of an image, making them highly dependent on critical visual evidence (highlighted in red in
Figure S4). Existing pruning methods select visual tokens only once before generation. If critical regions are omitted, they cannot be recovered during subsequent decoding. Moreover, unlike mathematical or other reasoning tasks, these questions are difficult to to be inferred directly from text alone. Consequently, models could repeatedly reassess candidate answers, ultimately leading to reasoning inflation and substantial accuracy degradation. Thus, the additional decoding overhead could even far exceed the computational savings achieved by pruning during prefill, resulting in significantly increased overall computational costs. Following your suggestion, we have added representative failure cases inLine 1040of the revised manuscript to further illustrate this phenomenon.