English
Related papers

Related papers: Weight Ensembling Improves Reasoning in Language M…

200 papers

Code-switching (CSW) is a common phenomenon among multilingual speakers where multiple languages are used in a single discourse or utterance. Mixed language utterances may still contain grammatical errors however, yet most existing Grammar…

Computation and Language · Computer Science 2024-08-13 Kelvin Wey Han Chan , Christopher Bryant , Li Nguyen , Andrew Caines , Zheng Yuan

Weight oscillation is an undesirable side effect of quantization-aware training, in which quantized weights frequently jump between two quantized levels, resulting in training instability and a sub-optimal final model. We discover that the…

Computer Vision and Pattern Recognition · Computer Science 2023-09-20 Shih-Yang Liu , Zechun Liu , Kwang-Ting Cheng

Supervised fine-tuning (SFT) of large language models can be viewed as an off-policy learning problem, where expert demonstrations come from a fixed behavior policy while training aims to optimize a target policy. Importance sampling is the…

Machine Learning · Computer Science 2025-09-22 Shiwan Zhao , Xuyang Zhao , Jiaming Zhou , Aobo Kong , Qicheng Li , Yong Qin

We introduce SIRI, Scaling Iterative Reinforcement Learning with Interleaved Compression, a simple yet effective RL approach for Large Reasoning Models (LRMs) that enables more efficient and accurate reasoning. Existing studies have…

Machine Learning · Computer Science 2025-09-30 Haoming Wen , Yushi Bai , Juanzi Li , Jie Tang

One critical challenge for large language models (LLMs) for making complex reasoning is their reliance on matching reasoning patterns from training data, instead of proactively selecting the most appropriate cognitive strategy to solve a…

Computation and Language · Computer Science 2025-03-18 Qin Liu , Wenxuan Zhou , Nan Xu , James Y. Huang , Fei Wang , Sheng Zhang , Hoifung Poon , Muhao Chen

Despite strong performance in many sequence-to-sequence tasks, autoregressive models trained with maximum likelihood estimation suffer from exposure bias, i.e. the discrepancy between the ground-truth prefixes used during training and the…

Computation and Language · Computer Science 2023-01-11 Michalis Korakakis , Andreas Vlachos

Weak-to-strong (W2S) generalization is a type of finetuning (FT) where a strong (large) student model is trained on pseudo-labels generated by a weak teacher. Surprisingly, W2S FT often outperforms the weak teacher. We seek to understand…

Machine Learning · Computer Science 2026-04-21 Yijun Dong , Yicheng Li , Yunai Li , Jason D. Lee , Qi Lei

Recent large language models (LLMs) are trained on diverse corpora and tasks, leading them to develop complementary strengths. Multi-agent debate (MAD) has emerged as a popular way to leverage these strengths for robust reasoning, though it…

Computer Vision and Pattern Recognition · Computer Science 2025-12-03 Anoop Cherian , River Doyle , Eyal Ben-Dov , Suhas Lohit , Kuan-Chuan Peng

Recent advances in large language models (LLMs), such as OpenAI-o1 and DeepSeek-R1, have demonstrated the effectiveness of test-time scaling, where extended reasoning processes substantially enhance model performance. Despite this, current…

Computation and Language · Computer Science 2025-03-26 Xiaoyu Tian , Sitong Zhao , Haotian Wang , Shuaiting Chen , Yunjie Ji , Yiping Peng , Han Zhao , Xiangang Li

Test-time training (TTT) methods explicitly update the weights of a model to adapt to the specific test instance, and they have found success in a variety of settings, including most recently language modeling and reasoning. To demystify…

Machine Learning · Computer Science 2026-02-24 Halil Alperen Gozeten , M. Emrullah Ildiz , Xuechen Zhang , Mahdi Soltanolkotabi , Marco Mondelli , Samet Oymak

Lung sepsis remains a significant concern in the Northeastern U.S., yet the national eICU Collaborative Database includes only a small number of patients from this region, highlighting underrepresentation. Understanding clinical variables…

Methodology · Statistics 2026-04-17 Subharup Guha , Mengqi Xu , Yi Li

Latent reasoning offers a computation-efficient alternative to Chain-of-Thought but often suffers from performance degradation due to distributional misalignment and ambiguous chain definitions. Ideally, latent reasoning should function as…

Computation and Language · Computer Science 2026-02-02 Jingcheng Deng , Liang Pang , Zihao Wei , Shicheng Xu , Zenghao Duan , Kun Xu , Yang Song , Huawei Shen , Xueqi Cheng

Reinforcement learning has become a powerful approach for enhancing large language model reasoning, but faces a fundamental dilemma: training on easy problems can cause overfitting and pass@k degradation, while training on hard problems…

Machine Learning · Computer Science 2026-05-04 Yangyi Fang , Jiaye Lin , Xiaoliang Fu , Cong Qin , Haolin Shi

Loss spikes, a phenomenon in which the loss value diverges suddenly, is a fundamental issue in the pre-training of large language models. This paper supposes that the non-uniformity of the norm of the parameters is one of the causes of loss…

Computation and Language · Computer Science 2024-10-08 Kosuke Nishida , Kyosuke Nishida , Kuniko Saito

Trained models are often composed with post-hoc transforms such as temperature scaling (TS), ensembling and stochastic weight averaging (SWA) to improve performance, robustness, uncertainty estimation, etc. However, such transforms are…

Machine Learning · Computer Science 2024-10-07 Rishabh Ranjan , Saurabh Garg , Mrigank Raman , Carlos Guestrin , Zachary Lipton

This study investigates the spatial reasoning capabilities of vision-language models (VLMs) through Chain-of-Thought (CoT) prompting and reinforcement learning. We begin by evaluating the impact of different prompting strategies and find…

Computer Vision and Pattern Recognition · Computer Science 2025-07-21 Binbin Ji , Siddharth Agrawal , Qiance Tang , Yvonne Wu

Language Models (LMs) exhibit two distinct mechanisms for knowledge acquisition: in-weights learning (i.e., encoding information within the model weights) and in-context learning (ICL). Although these two modes offer complementary…

Machine Learning · Computer Science 2026-04-03 Arslan Chaudhry , Sridhar Thiagarajan , Andrew Lampinen

Supervised fine-tuning (SFT) on visual instruction data often improves perceptual capabilities in vision-language models (VLMs) while degrading reasoning performance, creating a persistent reasoning tax during post-training. We investigate…

Computer Vision and Pattern Recognition · Computer Science 2026-03-30 Yiming Ren , Yujiu Yang , Junjie Wang

Many failures in deep continual and reinforcement learning are associated with increasing magnitudes of the weights, making them hard to change and potentially causing overfitting. While many methods address these learning failures, they…

Machine Learning · Computer Science 2024-07-03 Mohamed Elsayed , Qingfeng Lan , Clare Lyle , A. Rupam Mahmood

Few-shot learning is a challenging task that requires language models to generalize from limited examples. Large language models like GPT-3 and PaLM have made impressive progress in this area, but they still face difficulties in reasoning…

Computation and Language · Computer Science 2023-05-25 Yifei Li , Zeqi Lin , Shizhuo Zhang , Qiang Fu , Bei Chen , Jian-Guang Lou , Weizhu Chen