Just-Go-Parallel (Parallel Last (uni): EN→ZH)

The model repository for the "Parallel Last (uni): EN→ZH" setting of the following paper:

Just Go Parallel: Improving the Multilingual Capabilities of Large Language Models

Muhammad Reza Qorib, Junyi Li, and Hwee Tou Ng

The 63rd Annual Meeting of the Association for Computational Linguistics (to appear)

Paper: arXiv
Codebase: https://github.com/nusnlp/Just-Go-Parallel/

We use the architecture and tokenizer of TinyLlama v1.1. Please use transformers>=4.35.

Models

The main branch of the repository contains the best-performing model that was evaluated in the paper. Other checkpoints produced during training will also be hosted in this repository under different branch names (also called "revisions" in HuggingFace), with each branch name indicating the number of training steps.

No Parallel: nusnlp/JGP-No-Parallel
Multilingual: nusnlp/JGP-Multilingual
Parallel Non-Adjacent: nusnlp/JGP-Parallel-Non-Adjacent
Parallel First: nusnlp/JGP-Parallel-First
Parallel Distributed: nusnlp/JGP-Parallel-Distributed
Parallel Last (all): nusnlp/JGP-Parallel-Last-all
Parallel Last (uni):
- EN→ID: nusnlp/JGP-Parallel-Last-EN-ID
- ID→EN: nusnlp/JGP-Parallel-Last-ID-EN
- EN→ZH: nusnlp/JGP-Parallel-Last-EN-ZH
- ZH→EN: nusnlp/JGP-Parallel-Last-ZH-EN