11 papers · 1 filter
Multi-Mixer Models: Flexible Sequence Modeling with Shared Representations
Kevin Y. Li, Asher Trockman, Ananda Theertha Suresh +1
Softmax attention is the cornerstone of modern large language models, but its memory scales linearly and compute quadratically with sequence length. Linear recurrent models, such a…
CoDistill-GRPO: A Co-Distillation Recipe for Efficient Group Relative Policy Optimization
Soo Min Kwon, Ziteng Sun, Ananda Theertha Suresh +2
Group Relative Policy Optimization (GRPO) has emerged as a powerful algorithm for improving the reasoning capabilities of language models, but often fails to improve small models d…
CafeQ: Calibration-free Quantization via Learned Transformations and Adaptive Rounding
Ziteng Sun, Adrian Benton, Samuel Kushnir +4
Post-training quantization is an effective method for reducing the serving cost of large language models, where the standard approach is to use a round-to-nearest quantization leve…
Theoretical guarantees on the best-of-n alignment policy
Ahmad Beirami, Alekh Agarwal, Jonathan Berant +4
A simple and effective method for the inference-time alignment and scaling test-time compute of generative models is best-of- sampling, where samples are drawn from a refere…
Block Verification Accelerates Speculative Decoding
Ziteng Sun, Uri Mendlovic, Yaniv Leviathan +4
Speculative decoding is an effective method for lossless acceleration of large language models during inference. It uses a fast model to draft a block of tokens which are then veri…
Rate of Model Collapse in Recursive Training
Ananda Theertha Suresh, Andrew Thangaraj, Aditya Nanda Kishore Khandavally
Given the ease of creating synthetic data from machine learning models, new models can be potentially trained on synthetic data generated by previous models. This recursive trainin…