- Rating
- 3(+1)
- Confidence
- 4
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
The paper proposes FD-OPSD, which augments GRPO with self-distillation from a denser video view. Responses are generated using eight frames and scored by the same policy using 32 frames. Changes in token likelihoods, together with teacher confidence and trajectory reward, determine selective distillation weights. Experiments on two Qwen backbones and six benchmarks show modest average improvements over the strongest RL baselines.
Critical strengths
-
Practical use of dense observations. Dense view scoring provides an auxiliary signal without a separate teacher or dense autoregressive rollouts, and introduces no additional inference cost.
-
Reasonably broad empirical evaluation. Experiments cover two backbones, six benchmarks, and multiple frame budgets, with component ablations and training-time comparisons.
Critical weaknesses
-
Limited and uncertain benefit of OPSD. In my own experiments, OPSD has not yielded useful improvements, which makes me skeptical of its practical value here.
-
The reported average gains over the strongest RL baseline are only 0.8–1.3 points. Given the finite evaluation sets and the variability I have observed in such evaluations, these gains are not convincing.
-
Baseline performance appears too low. Qwen models often perform better with direct answer prompting than with CoT. I would expect Qwen3-VL-4B to score above 60 on VideoMME with 32 frames, whereas the paper reports only 53.6. The evaluation appears to use CoT, raising concerns that the prompting setup understates baseline performance and inflates the apparent improvement from post-training.
Questions for authors
-
What motivates the use of OPSD?
-
Is CoT required during evaluation? Please specify the prompts, decoding settings, and answer extraction, and compare direct answer and CoT evaluation for both the instruction checkpoints and FD-OPSD under identical frame and pixel budgets.