12 papers
EMO: A Closed-Loop Test-Time Computation Framework via Alternating Context Optimization and Logits Steering
Wei Xia, Yehui Tang, Zhi-Hong Deng
Test-time context optimization adapts frozen language models through textual feedback, but ordinary conditional generation provides no separate control over how strongly to reinfor…
The World Won't Wait: Evaluating and Teaching Agents Under a Live Clock
Ningxin Pan, Hanyu Li, Yeqing Shen +3
Large language model (LLM) agents can solve complex interactive tasks, yet reasoning and tool calls take time. In continuously changing environments, an otherwise correct action ma…
What Matters in Designing World Action Models: An Empirical Study
Chao Tang, Haoqing Wang, Zilang Cen +9
World Action Models (WAMs) have emerged as a promising paradigm for generalizable robot control. Despite the growing number of WAM systems, existing works often introduce unified s…
ComplexityWorld: Benchmarking Vision-Language Models on Verifiable Visual Decision Making
Ningxin Pan, Hanyu Li, Yehui Tang
Vision-language models (VLMs) have made rapid progress in visual perception and increasingly support real-world tasks that depend on images. Many such tasks, however, require more…
FaithEyes: Towards Faithful Tool Use via Multi-Agent Process-Image Self-Verification
Haoqing Wang, Xingrun Xing, Wei Xia +3
Agentic vision-language models (VLMs), which interleave textual reasoning with explicit tool calls such as cropping and code-based image manipulation, have emerged as a compelling…
Trust Region On-Policy Distillation
Xingrun Xing, Haoqing Wang, Boyan Gao +2
On-Policy Distillation (OPD) is a fundamental technique for efficient post-training of large language models (LLMs), with broad applications in agent learning, multi-task enhanceme…
When Should LLMs Reason? Adaptive Inference Routing via Entropy Dynamics
Wei Xia, Haoqing Wang, Zhi-Hong Deng +1
Chain-of-thought (CoT) reasoning can improve LLM performance, but its benefits vary across tasks, models, and individual queries. On some tasks, CoT provides marginal or negative a…
MemTrain: Self-Supervised Context Memory Training
Ziheng Li, Xingrun Xing, Haoqing Wang +2
Memory is an indispensable capability for long-horizon LLM agents, enabling them to preserve and utilize information accumulated across extended interactions. Existing memory-agent…
The MiniMax-M2 Series: Mini Activations Unleashing Max Real-World Intelligence
Aili Chen, Aonian Li, Baichuan Zhou +215
We introduce the MiniMax-M2 series, a family of Mixture-of-Experts language models built around the principle that mini activations can unleash maximum real-world intelligence. The…
LiveClawBench: Benchmarking LLM Agents on Complex, Real-World Assistant Tasks
Xiang Long, Li Du, Yilong Xu +11
OpenClaw-style personal assistants extend LLM agents from isolated tool use to open-ended, stateful, and personalized software environments. Evaluating these assistants is fundamen…
MeKi: Memory-based Expert Knowledge Injection for Efficient LLM Scaling
Ning Ding, Fangcheng Liu, Kyungrae Kim +4
Scaling Large Language Models (LLMs) typically relies on increasing the number of parameters or test-time computations to boost performance. However, these strategies are impractic…
To Mix or To Merge: Toward Multi-Domain Reinforcement Learning for Large Language Models
Haoqing Wang, Xiang Long, Ziheng Li +3
Reinforcement Learning with Verifiable Rewards (RLVR) plays a key role in stimulating the explicit reasoning capability of Large Language Models (LLMs). We can achieve expert-level…