LMFlow: An Extensible Toolkit for Finetuning and Inference of Large Foundation Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Diao, Shizhe, Pan, Rui, Dong, Hanze, Shum, Ka Shun, Zhang, Jipeng, Xiong, Wei, Zhang, Tong
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913340473737216
author Diao, Shizhe
Pan, Rui
Dong, Hanze
Shum, Ka Shun
Zhang, Jipeng
Xiong, Wei
Zhang, Tong
author_facet Diao, Shizhe
Pan, Rui
Dong, Hanze
Shum, Ka Shun
Zhang, Jipeng
Xiong, Wei
Zhang, Tong
contents Foundation models have demonstrated a great ability to achieve general human-level intelligence far beyond traditional approaches. As the technique keeps attracting attention from the AI community, an increasing number of foundation models are becoming publicly accessible. However, a significant shortcoming of most of these models lies in their performance in specialized-domain and task-specific applications, necessitating domain- and task-aware fine-tuning to develop effective scientific language models. As the number of available foundation models and specialized tasks keeps growing, the job of training scientific language models becomes highly nontrivial. In this paper, we initiate steps to tackle this issue. We introduce an extensible and lightweight toolkit, LMFlow, which aims to simplify the domain- and task-aware finetuning of general foundation models. LMFlow offers a complete finetuning workflow for a foundation model to support specialized training with limited computing resources. Furthermore, it supports continuous pretraining, instruction tuning, parameter-efficient finetuning, alignment tuning, inference acceleration, long context generalization, model customization, and even multimodal finetuning, along with carefully designed and extensible APIs. This toolkit has been thoroughly tested and is available at https://github.com/OptimalScale/LMFlow.
format Preprint
id arxiv_https___arxiv_org_abs_2306_12420
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle LMFlow: An Extensible Toolkit for Finetuning and Inference of Large Foundation Models
Diao, Shizhe
Pan, Rui
Dong, Hanze
Shum, Ka Shun
Zhang, Jipeng
Xiong, Wei
Zhang, Tong
Computation and Language
Artificial Intelligence
Foundation models have demonstrated a great ability to achieve general human-level intelligence far beyond traditional approaches. As the technique keeps attracting attention from the AI community, an increasing number of foundation models are becoming publicly accessible. However, a significant shortcoming of most of these models lies in their performance in specialized-domain and task-specific applications, necessitating domain- and task-aware fine-tuning to develop effective scientific language models. As the number of available foundation models and specialized tasks keeps growing, the job of training scientific language models becomes highly nontrivial. In this paper, we initiate steps to tackle this issue. We introduce an extensible and lightweight toolkit, LMFlow, which aims to simplify the domain- and task-aware finetuning of general foundation models. LMFlow offers a complete finetuning workflow for a foundation model to support specialized training with limited computing resources. Furthermore, it supports continuous pretraining, instruction tuning, parameter-efficient finetuning, alignment tuning, inference acceleration, long context generalization, model customization, and even multimodal finetuning, along with carefully designed and extensible APIs. This toolkit has been thoroughly tested and is available at https://github.com/OptimalScale/LMFlow.
title LMFlow: An Extensible Toolkit for Finetuning and Inference of Large Foundation Models
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2306.12420