홈 › Papers

Proofread: Fixes All Errors with One Tap

2024-06-06 · Renjie Liu, Yanxiang Zhang, Yun Zhu, Haicheng Sun, Yuanbo Zhang, Michael Xuelin Huang, Shanqing Cai, Lei Meng, Shumin Zhai

The impressive capabilities in Large Language Models (LLMs) provide a powerful approach to reimagine users' typing experience. This paper demonstrates Proofread, a novel Gboard feature powered by a server-side LLM in Gboard, enabling seamless sentence-level and paragraph-level corrections with a single tap. We describe the complete system in this paper, from data generation, metrics design to model tuning and deployment. To obtain models with sufficient quality, we implement a careful data synthetic pipeline tailored to online use cases, design multifaceted metrics, employ a two-stage tuning approach to acquire the dedicated LLM for the feature: the Supervised Fine Tuning (SFT) for foundational quality, followed by the Reinforcement Learning (RL) tuning approach for targeted refinement. Specifically, we find sequential tuning on Rewrite and proofread tasks yields the best quality in SFT stage, and propose global and direct rewards in the RL tuning stage to seek further improvement. Extensive experiments on a human-labeled golden set showed our tuned PaLM2-XS model achieved 85.56\% good ratio. We launched the feature to Pixel 8 devices by serving the model on TPU v5 in Google Cloud, with thousands of daily active users. Serving latency was significantly reduced by quantization, bucket inference, text segmentation, and speculative decoding. Our demo could be seen in \href{https://youtu.be/4ZdcuiwFU7I}{Youtube}.

📄 PDF Abstract BibTeX arXiv:2406.04523

Code (0)

등록된 구현이 없습니다.

Tasks

AllQuantization Reinforcement Learning (RL)SentenceText Segmentation

Methods 이 논문이 사용한 방법론

SET Dynamic Sparse Training method where weight mask is updated randomly periodically

SFT Shrink and Fine-Tune, or SFT, is a type of distillation that avoids explicit distillation by copying parameters to a student student model and then fine-tuning.…

VICE: Visual Identification and Correction of Neural Circuit Errors

2021-05-14 · Felix Gonda, Xueying Wang, Johanna Beyer, Markus Hadwiger 외

A connectivity graph of neurons at the resolution of single synapses provides scientists with a tool for understanding the nervous system in health and disease. Recent advances in automatic image segmentation and synapse…

Clustering Image Segmentation Segmentation Semantic Segmentation

Guided Proofreading of Automatic Segmentations for Connectomics

2017-04-04 · CVPR 2018 6 · Daniel Haehn, Verena Kaynig, James Tompkin, Jeff W. Lichtman 외

Automatic cell image segmentation methods in connectomics produce merge and split errors, which require correction through proofreading. Previous research has identified the visual search for these errors as the bottlene…

Image Segmentation Segmentation Semantic Segmentation

RLCorrector: Reinforced Proofreading for Cell-level Microscopy Image Segmentation

2021-06-10 · Khoa Tuan Nguyen, Ganghee Jang, Tran Anh Tuan, Won-Ki Jeong

Segmentation of nanoscale electron microscopy (EM) images is crucial but still challenging in connectomics research. One reason for this is that none of the existing segmentation methods are error-free, so they require p…

Image Segmentation reinforcement-learning Reinforcement Learning Reinforcement Learning (RL)+2

Thermodynamics of Error Correction

2015-12-11

Information processing at the molecular scale is limited by thermal fluctuations. This can cause undesired consequences in copying information since thermal noise can lead to errors that can compromise the functionality …

ConnectomeBench2: A Unified Benchmark for Automated Connectomic Proofreading

2026-06-19 · Jeff Brown, Tim Farkas, Gleb Razgar, Edward S. Boyden arxiv

Proofreading--correcting segmentation errors in 3D brain reconstructions--is the rate-limiting step in synapse-resolution connectomics. We release ConnectomeBench2, a unified multi-species dataset of over 716,485 expert-…

Active Learning