Mid-training equips pretrained large language models with specialized and reasoning capabilities, but the returns of this stage are bounded since additional serial compute yields little further downstream improvement and can even degrade some capabilities, which places a practical ceiling on how much compute mid-traini...
Zhehao Huang, Changxin Tian, Qingyuan Yang et al.· 0 citations
Recent approaches to reinforcement learning (RL) post-training for large language models increasingly remove the critic to reduce training instability and memory overhead. Even where a critic is trained, it is discarded once training ends, although it has learned to predict outcomes. We revisit this trend and show that...
Hong-Yang Li, Xiao Li, Caesar Wu et al.· 0 citations
These stages establish a constructive post-training path from English-pivoted capability to multilingual reasoning that is reliably delivered, and jointly track correctness, language adherence, termination, and delivery efficiency.
Hong-Yang Li, Xiao Li, Caesar Wu et al.· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
It is shown that useful explanations can extend beyond the model a verbalizer was trained to describe and show that pretrained verbalizers recover words that models have learned to conceal through fine-tuning, without additional verbalizer training.
Federico Torrielli, Gianluca Barmina, Andrea Blasi Núñez et al.· 0 citations
Knowledge concept tagging aims to assign specific concept or topic labels to educational content, which is essential for both educators and learners in traditional and online teaching practices. Recent work has explored large language models (LLMs) for this task, achieving promising performance. However, LLMs still str...
Zhiwei Yang, Jiahua Yang, Huiru Lin et al.· 0 citations
These measurements turn correlation from a post-hoc warning into an auditable allocation decision and turn correlation from a post-hoc warning into an auditable allocation decision.
Miao-Bo Hu, Shu-Hao Hu, Xiao-Bo Guo et al.· 0 citations
Large language models (LLMs) excel at token-level generation but may learn undesirable abstract semantics and lack comprehensive perception. LLM-JEPA mitigates this by aligning different views of the same underlying knowledge via a joint-embedding predictive architecture (JEPA). However, strong alignment does not neces...
BaLEEN (Biasing with Latent Encoded Entities), a lightweight, hypernetwork-based framework for dynamic contextual adaptation without fine-tuning the underlying ASR model, is proposed.
A bilingual Moshi-style model is trained that sustains coherent multi-party English-Chinese conversations over extended durations and substantially outperforms open-source baselines including Moshi, MiniCPM-o-4.5, and Qwen3-Omni-30B-A3B-Instruct on MultiTalkBench.
Ke Wang, Houxing Ren, Zi-Mu Lu et al.· 0 citations
Reflection on Visual Evidence (ReVuE), an on-policy distillation method for visual agents that translates trajectory-level evidence diagnosis into targeted token-level supervision, guiding students to improve their visual evidence acquisition and reasoning.
Shaohang Wei, Feifan Song, Guangyue Peng et al.· 0 citations
Chinese Semantic Error Correction (CSEC) targets semantic errors in Chinese text, which are typically more subtle and complex than spelling and grammatical errors but remain relatively underexplored. Existing LLM-based approaches face two recurring obstacles in this task: over-correction, and unclear interaction betwee...
A new method, called CW-Net, translates the reasoning process of an autonomous vehicle’s AI system into understandable concepts that explain its behavior.
MIT News · Artificial Intelligence· news.mit.eduAug 31, 2026
With millions of users across the world, Julia has been used to conduct cutting-edge research and to design new drugs, jet engines, heat pumps, and more.
