English
Related papers

Related papers: The Optimal Token Baseline: Variance Reduction for…

200 papers

While Large Language Models (LLMs) demonstrate exceptional performance in surface-level text generation, their nature in handling complex multi-step reasoning tasks often remains one of ``statistical fitting'' rather than systematic logical…

Machine Learning · Computer Science 2026-01-27 Lianlei Shan , Han Chen , Yixuan Wang , Zhenjie Liu , Wei Li

Large language models pretrained on general-domain corpora often exhibit tokenization inefficiencies when applied to specialized domains. Although continual pretraining for domain adaptation partially alleviate performance degradation, it…

Computation and Language · Computer Science 2026-05-19 Gunjan Balde , Soumyadeep Roy , Mainack Mondal , Niloy Ganguly

Deep neural network training involves both forward propagation (from features through logits to loss) and backward propagation (from loss through gradients to parameter updates). While perturbations along the forward chain, including…

Machine Learning · Computer Science 2026-05-29 Hua Li

Sequential decision making in the real world often requires finding a good balance of conflicting objectives. In general, there exist a plethora of Pareto-optimal policies that embody different patterns of compromises between objectives,…

Machine Learning · Computer Science 2024-10-08 Takuya Kanazawa , Chetan Gupta

This work studies a Reinforcement Learning (RL) problem in which we are given a set of trajectories collected with K baseline policies. Each of these policies can be quite suboptimal in isolation, and have strong performance in…

Machine Learning · Computer Science 2024-03-29 Teodor V. Marinov , Alekh Agarwal , Mircea Trofin

Aligning Large Language Models (LLMs) with human preferences is critical, yet traditional fine-tuning methods are computationally expensive and inflexible. While test-time alignment offers a promising alternative, existing approaches often…

Artificial Intelligence · Computer Science 2026-01-16 Tiesunlong Shen , Rui Mao , Jin Wang , Heming Sun , Jian Zhang , Xuejie Zhang , Erik Cambria

The prevalent use of Byte Pair Encoding (BPE) in Large Language Models (LLMs) facilitates robust handling of subword units and avoids issues of out-of-vocabulary words. Despite its success, a critical challenge persists: long tokens, rich…

Computation and Language · Computer Science 2024-11-11 Haoran Lian , Yizhe Xiong , Zijia Lin , Jianwei Niu , Shasha Mo , Hui Chen , Peng Liu , Guiguang Ding

Preference learning in Large Language Models (LLMs) has advanced significantly, yet existing methods remain limited by modest performance gains, high computational costs, hyperparameter sensitivity, and insufficient modeling of global…

Computation and Language · Computer Science 2026-04-03 Liang Zhu , Yuelin Bai , Xiankun Ren , Jiaxi Yang , Lei Zhang , Feiteng Fang , Hamid Alinejad-Rokny , Minghuan Tan , Min Yang

Safety post-training can improve the harmfulness and policy compliance of Large Language Models (LLMs), but it may also reduce general utility, a phenomenon often described as the \emph{alignment tax}. We study this trade-off through the…

Machine Learning · Computer Science 2026-05-13 Guanglong Sun , Siyuan Zhang , Liyuan Wang , Jun Zhu , Hang Su , Yi Zhong

Correctness-based Reinforcement Learning with Verifiable Rewards (RLVR) trains language models from binary feedback on sampled outputs, but the objective optimized in expectation and the stochastic update geometry induced by finite rollout…

Machine Learning · Computer Science 2026-05-29 Yifu Zheng

Large Language Models (LLMs) have achieved remarkable success on reasoning benchmarks through Reinforcement Learning with Verifiable Rewards (RLVR), excelling at tasks such as math, coding, logic, and puzzles. However, existing benchmarks…

Artificial Intelligence · Computer Science 2026-05-12 Xiaozhe Li , Xinyu Fang , Shengyuan Ding , Yang Li , Linyang Li , Haodong Duan , Qingwen Liu , Kai Chen

Model-based reinforcement learning (RL) offers a compelling approach to offline RL by enabling value learning on imagined on-policy trajectories. However, it often suffers from compounding errors due to repeated model inference on…

Machine Learning · Computer Science 2026-05-18 Hojun Chung , Junseo Lee , Songhwai Oh

Previous language model pre-training methods have uniformly applied a next-token prediction loss to all training tokens. Challenging this norm, we posit that "9l training". Our initial analysis examines token-level training dynamics of…

Computation and Language · Computer Science 2025-01-09 Zhenghao Lin , Zhibin Gou , Yeyun Gong , Xiao Liu , Yelong Shen , Ruochen Xu , Chen Lin , Yujiu Yang , Jian Jiao , Nan Duan , Weizhu Chen

Bandit and reinforcement learning (RL) problems can often be framed as optimization problems where the goal is to maximize average performance while having access only to stochastic estimates of the true gradient. Traditionally, stochastic…

Machine Learning · Computer Science 2021-02-22 Wesley Chung , Valentin Thomas , Marlos C. Machado , Nicolas Le Roux

Distributed training and increasing the gradient update frequency are practical strategies to accelerate learning and improve performance, but both exacerbate a central challenge: \textit{policy lag}, which is the mismatch between the…

Reinforcement learning (RL) commonly relies on scalar rewards with limited ability to express temporal, conditional, or safety-critical goals, and can lead to reward hacking. Temporal logic expressible via the more general class of…

Artificial Intelligence · Computer Science 2025-11-26 Dominik Wagner , Leon Witzman , Luke Ong

Large language models (LLMs) such as GPT-5 and Gemini 3 have pushed the frontier of automated reasoning and code generation. Yet current benchmarks emphasize accuracy and output quality, neglecting a critical dimension: efficiency of token…

Computation and Language · Computer Science 2026-02-25 Zheng Du , Hao Kang , Song Han , Tushar Krishna , Ligeng Zhu

Group Relative Policy Optimisation (GRPO) enhances large language models by estimating advantages across a group of sampled trajectories. However, mapping these trajectory-level advantages to policy updates requires aggregating token-level…

Cross-domain task-oriented dialogue requires reasoning over implicit and explicit feasibility constraints while planning long-horizon, multi-turn actions. Large language models (LLMs) can infer such constraints but are unreliable over long…

Computation and Language · Computer Science 2026-04-28 Yangyang Zhao , Linfan Dai , Li Cai , Bowen Xing , Libo Qin

Advanced large-scale neural language models have led to significant success in many language generation tasks. However, the most commonly used training objective, Maximum Likelihood Estimation (MLE), has been shown problematic, where the…

Computation and Language · Computer Science 2021-06-15 Xiang Lin , Simeng Han , Shafiq Joty