Indexxlim

Notes from a slow studio, collected every week.

A curated log of stories, sketches, and quiet observations. Browse the journal, explore categories, and settle into a calmer pace.

5Entries

5Collections

4Updates / mo

Featured Story

Efficient Memory Management for Large Language Model Serving with PagedAttention

Jisu Lim·June 20, 2023·Large Language Models, Memory Management, PagedAttention, vLLM·16 min read

Jisu Lim·June 1, 2023·Model Quantization, Large Language Models, Model Compression, Model Acceleration·11 min read

Abstractact

Jisu Lim·March 6, 2023·Speech Recognition, Multilingual, Google, Universal Speech Model·9 min read

Abstract and Introduction

Jisu Lim·February 9, 2023·Natural Language Processing, Language Models, Tools, API·3 min read

Abstract

Jisu Lim·January 1, 2023·Natural Language Processing, Pre-training, Sequence-to-Sequence, BART·6 min read

We present BART, a denoising autoencoder for pretraining sequence-to-sequence models.