PROGRESSLM: Towards Progress Reasoning in Vision-Language Models
This addresses the challenge of enabling AI systems to reason about long-horizon dynamics in tasks like robotics or video analysis, though it is incremental as it builds on existing VLM capabilities.
The paper tackled the problem of whether vision-language models can estimate task progress from partial observations, finding that most models are not yet ready for this task, but a training-based approach called ProgressLM-3B achieved consistent improvements even with disjoint training and evaluation tasks.
Estimating task progress requires reasoning over long-horizon dynamics rather than recognizing static visual content. While modern Vision-Language Models (VLMs) excel at describing what is visible, it remains unclear whether they can infer how far a task has progressed from partial observations. To this end, we introduce Progress-Bench, a benchmark for systematically evaluating progress reasoning in VLMs. Beyond benchmarking, we further explore a human-inspired two-stage progress reasoning paradigm through both training-free prompting and training-based approach based on curated dataset ProgressLM-45K. Experiments on 14 VLMs show that most models are not yet ready for task progress estimation, exhibiting sensitivity to demonstration modality and viewpoint changes, as well as poor handling of unanswerable cases. While training-free prompting that enforces structured progress reasoning yields limited and model-dependent gains, the training-based ProgressLM-3B achieves consistent improvements even at a small model scale, despite being trained on a task set fully disjoint from the evaluation tasks. Further analyses reveal characteristic error patterns and clarify when and why progress reasoning succeeds or fails.