Release notes from TextBox

TextBox 2.0 Release

2022-12-28T02:06:22Z

TextBox 2.0 is an up-to-date text generation library based on Python and PyTorch focusing on building a unified and standardized pipeline for applying pre-trained language models to text generation:

From a task perspective, we consider 13 common text generation tasks such as translation, story generation, and style transfer, and their corresponding 83 widely-used datasets.
From a model perspective, we incorporate 47 pre-trained language models/modules covering the categories of general, translation, Chinese, dialogue, controllable, distilled, prompting, and lightweight models (modules).
From a training perspective, we support 4 pre-training objectives and 4 efficient and robust training strategies, such as distributed data parallel and efficient generation.

Compared with the previous version of TextBox, this extension mainly focuses on building a unified, flexible, and standardized framework for better supporting PLM-based text generation models. There are three advantages of TextBox 2.0:

It is a significant innovation focusing on comprehensive tasks and PLMs.
It is designed to be unified in implementation and interface.
It can faithfully reproduce the results reported in existing work.

TextBox v0.2.1

2021-04-15T13:39:28Z

TextBox v0.2.1 Release Notes

The TextBox v0.2.1 release includes a number of wonderful new features, some bug fixes and code refactor. A few of the highlights include:

We add 6 new models: HRED, CVAE, T5, ProphetNet, Context2Seq and Attribute2Seq.
We add 3 new datasets: Persona Chat for dialog system, Amazon Electronic for attribute to text generation and Chinese Classical Poetry Corpus for poem generation.
We support Distributed Data Parallel (DDP) for training with multiple GPUs conveniently.
We refactor the codes of pretrained language models (PLMs) for improving performance.
We refactor the dataset and dataloader to provide unified and convenient interface.
We unify and simplify the generate function for each model.
We unify the config parameters of different models and datasets.

TextBox v0.1.5

2021-01-11T12:54:55Z

TextBox is an open-source library for building text generation system. It is developed based on Python and PyTorch.