12 papers · 1 filter
Fast and Expressive Multi-Byte Prediction with Probabilistic Circuits
Andreas Grivas, Lorenzo Loconte, Emile van Krieken +6
Multi-token prediction (MTP) is a prominent strategy to significantly speed up generation in large language models (LLMs), especially in byte-level LLMs, which are tokeniser-free b…
Blending Supervised and Reinforcement Fine-Tuning with Prefix Sampling
Zeyu Huang, Tianhao Cheng, Zihan Qiu +4
Existing LLMs-post-training techniques are broadly categorized into supervised fine-tuning (SFT) and reinforcement fine-tuning (RFT). Each paradigm presents a distinct trade-off: (…
The Cancellation Hypothesis in Critic-Free RL: From Outcome Rewards to Token Credits
Tianhao Cheng, Zeyu Huang, Zihan Qiu +5
A commonly accepted explanation of critic-free RL for LLMs, based on sequence-level rewards, is that it reinforces successful rollouts with a positive advantage while penalizing fa…
AdaSplash-2: Faster Differentiable Sparse Attention
Nuno Gonçalves, Hugo Pitorro, Vlad Niculae +4
Sparse attention has been proposed as a way to alleviate the quadratic cost of transformers, a central bottleneck in long-context training. A promising line of work is -entmax…
Adapting Time Series Foundation Models through Data Mixtures
Thomas L. Lee, Edoardo M. Ponti, Amos Storkey
Time series foundation models (TSFMs) have become increasingly popular for zero-shot forecasting. However, for a new time series domain not fully covered by the pretraining set, pe…
Self-Improving World Modelling with Latent Actions
Yifu Qiu, Zheng Zhao, Waylon Li +4
Internal modelling of the world -- predicting transitions between previous states and next states under actions -- is essential to reasoning and planning for LLMs and V…