Reinforcement learning with verifiable rewards (RLVR) improves multimodal reasoning, but uniformly assigning a single response-level advantage to all tokens overlooks the distinct roles of visual observations and subsequent reasoning. Visual affinity alone is insufficient: predictive states may align with both image an...
Yu-Hang Han, Ding-Ning Liu, Yu-Jie Chen et al.· 0 citations
Text-to-video (T2V) models are highly sensitive to prompt specificity, while iterative prompt optimization typically improves generation quality only through repeated video generation and vision-language model evaluation at inference time. We propose TIDE, a framework that distills this costly search process into an in...
A central challenge in embodied recursive self-improvement is enabling agents to convert interaction history into capabilities that persist beyond one world and transfer to later tasks. A saved memory or successful final action does not by itself establish that a reusable capability was acquired; evaluation must separa...
Jing-Wei Song, Zhi-Sheng Chen, Bing-Fan Zeng et al.· 1 citation
Long-term agents face growing storage demands as they accumulate experience. World models capture reusable regularities that can reduce the information stored for each experience. We formulate the problem of memory allocation conditioned on a world model and introduce MemoWM, a framework that uses shared predictions to...
Bing-Fan Zeng, Zhi-Sheng Chen, Chen-Bo Sang et al.· 0 citations
ESCD aggregates prefix-related teacher events and supervises the total probability of byte-compatible one-step student completions, avoiding tokenizer-dependent probability splits among individual tokens, and support event entry and event completion as complementary supervision targets for cross-tokenizer knowledge tra...
ToolVision is introduced, a model that improves over its base on all seven main benchmarks, surpasses Thyme-7B, CodeVision-8B, and CodeDance-7B on all three high-resolution benchmarks, and outperforms Qwen3-VL-32B-Thinking on V* and HRBench 8K.
Delin Mao, Cheng-Hao Sun, Jing-Wei Song et al.· 1 citation
Experiments on long-horizon embodied and web-agent benchmarks show that MemPrism consistently improves the task performance, especially as trajectories become longer, while reducing memory token consumption.
Zhi-Sheng Chen, Bingfan Zeng, Bangde Cao et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.