6 papers
Spend Your Rollouts Where It Counts: Rollout Allocation for Group-Based RL Post-Training
Woojeong Kim, Ziyi Yang, Jing Nathan Yan +1
Reinforcement learning (RL) is the dominant paradigm for post-training large language models. However, in the online, on-policy setting, rollout generation dominates the computatio…
The Efficiency Gap in Byte Modeling
Celine Lee, Jing Nathan Yan, Chen Liang +9
Modern language models have historically relied on two dominant design choices: subword tokenization and autoregressive (AR) ordering. These design decisions bake in priors that di…
CrowdVLM-R1: Expanding R1 Ability to Vision Language Model for Crowd Counting using Fuzzy Group Relative Policy Reward
Zhiqiang Wang, Pengbin Feng, Yanbin Lin +4
We propose Fuzzy Group Relative Policy Reward (FGRPR), a novel framework that integrates Group Relative Policy Optimization (GRPO) with a fuzzy reward function to enhance learning…
OverFill: Two-Stage Models for Efficient Language Model Decoding
Woojeong Kim, Junxiong Wang, Jing Nathan Yan +2
Large language models (LLMs) excel across diverse tasks but face significant deployment challenges due to high inference costs. LLM inference comprises prefill (compute-bound) and…
State Space Models are Strong Text Rerankers
Zhichao Xu, Jinghua Yan, Ashim Gupta +1
Transformers dominate NLP and IR; but their inference inefficiencies and challenges in extrapolating to longer contexts have sparked interest in alternative model architectures. Am…
TopV: Compatible Token Pruning with Inference Time Optimization for Fast and Low-Memory Multimodal Vision Language Model
Cheng Yang, Yang Sui, Jinqi Xiao +8
Vision-Language Models (VLMs) demand substantial computational resources during inference, largely due to the extensive visual input tokens for representing visual information. Pre…