REVIEW 12 cited by
Unhackable Temporal Rewarding for Scalable Video MLLMs
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Unhackable Temporal Rewarding for Scalable Video MLLMs
read the original abstract
In the pursuit of superior video-processing MLLMs, we have encountered a perplexing paradox: the "anti-scaling law", where more data and larger models lead to worse performance. This study unmasks the culprit: "temporal hacking", a phenomenon where models shortcut by fixating on select frames, missing the full video narrative. In this work, we systematically establish a comprehensive theory of temporal hacking, defining it from a reinforcement learning perspective, introducing the Temporal Perplexity (TPL) score to assess this misalignment, and proposing the Unhackable Temporal Rewarding (UTR) framework to mitigate the temporal hacking. Both theoretically and empirically, TPL proves to be a reliable indicator of temporal modeling quality, correlating strongly with frame activation patterns. Extensive experiments reveal that UTR not only counters temporal hacking but significantly elevates video comprehension capabilities. This work not only advances video-AI systems but also illuminates the critical importance of aligning proxy rewards with true objectives in MLLM development.
Forward citations
Cited by 12 Pith papers
-
MentalThink: Shaping Thoughts in Mental SVG World
MLLMs that generate and render SVG sketches as multi-turn intermediate reasoning steps reach 55.1% on VSIBench and 76.0% on MindCube, far above the Qwen2.5-VL-7B backbone.
-
VIDEOP2R: Video Understanding from Perception to Reasoning
VideoP2R separates perception and reasoning in a process-aware RFT pipeline with a new CoT dataset and PA-GRPO rewards, reaching SOTA on six of seven video benchmarks.
-
Video-R1: Reinforcing Video Reasoning in MLLMs
Video-R1 uses temporal-aware RL and mixed datasets to boost video reasoning in MLLMs, with a 7B model reaching 37.1% on VSI-Bench and surpassing GPT-4o.
-
Latent Visual Cache for Video Reasoning
A recurrent latent visual cache in the decoder, trained with contrastive key-frame alignment and a vision-grounded GRPO reward, improves video reasoning accuracy while shortening answers.
-
CARE: Competence-Aware Reward Shaping for Adaptive Reasoning Length in Video-MLLMs
CARE uses exponential moving average competence estimates to progressively shift RL rewards from exploration-oriented long reasoning to efficiency-oriented concise reasoning in video-MLLMs, with batch normalization an...
-
Reasoning as Intersection: Consensus-Frame Alignment for Visual Focus in Video-MLLMs
CF-GRPO creates a consensus frame prior from intrinsic video cues and aligns it with model frame-use scores via a reward signal to enable evidence-aware reasoning in Video-MLLMs without temporal annotations.
-
Learning Spatiotemporal Sensitivity in Video LLMs via Counterfactual Reinforcement Learning
CRPO applies counterfactual videos and a cross-branch relation reward in RL post-training to reduce shortcut reliance in Video LLMs, with gains shown on the new DyBench paired benchmark.
-
Injecting Distributional Awareness into MLLMs via Reinforcement Learning for Deep Imbalanced Regression
A plug-and-play RL method adds batch-level distributional supervision via CCC rewards to reduce regression-to-the-mean in MLLMs on imbalanced regression benchmarks.
-
Injecting Distributional Awareness into MLLMs via Reinforcement Learning for Deep Imbalanced Regression
A Group Relative Policy Optimization framework with concordance correlation coefficient rewards improves MLLM regression accuracy on long-tailed distributions, especially in medium- and few-shot regimes, without model...
-
Beyond Perceptual Shortcuts: Causal-Inspired Debiasing Optimization for Generalizable Video Reasoning in Lightweight MLLMs
VideoThinker improves lightweight MLLM video reasoning by creating a bias model to capture shortcuts and applying causal debiasing policy optimization to push away from them, achieving SOTA efficiency with minimal data.
-
Video Reasoning without Training
An entropy-guided, inference-time value-cache controller improves video reasoning accuracy and cuts output tokens versus RL-trained baselines.
-
MUSEG: Reinforcing Video Temporal Understanding via Timestamp-Aware Multi-Segment Grounding
MUSEG applies timestamp-aware multi-segment grounding with a phased-reward RL recipe to boost temporal grounding and time-sensitive video QA performance in MLLMs.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.