English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Grammatical error correction aims to correct ungrammatical sentencesautomatically. Recently, some work has demonstrated the excellent capabilitiesof closed-source Large Language Models (LLMs, e.g., ChatGPT) in grammaticalerror correction. However, the potential of open-source LLMs remainsunexplored. In this paper, we introduced GrammarGPT, an open-source LLM, topreliminary explore its potential for native Chinese grammatical errorcorrection. The core recipe of GrammarGPT is to leverage the hybrid dataset ofChatGPT-generated and human-annotated. For grammatical errors with clues, weproposed a heuristic method to guide ChatGPT to generate ungrammaticalsentences by providing those clues. For grammatical errors without clues, wecollected ungrammatical sentences from publicly available websites and manuallycorrected them. In addition, we employed an error-invariant augmentation methodto enhance the ability of the model to correct native Chinese grammaticalerrors. We ultimately constructed about 1k parallel data and utilized thesedata to fine-tune open-source LLMs (e.g., Phoenix, released by The ChineseUniversity of Hong Kong, Shenzhen) with instruction tuning. The experimentalresults show that GrammarGPT outperforms the existing SOTA systemsignificantly. Although model parameters are 20x larger than the SOTA baseline,the required amount of data for instruction tuning is 1200x smaller,illustrating the potential of open-source LLMs on native CGEC. Our GrammarGPTranks $3^{rd}$ on NLPCC2023 SharedTask1, demonstrating our approach'seffectiveness. The code and data are available at\url{https://github.com/FreedomIntelligence/GrammarGPT}.

GrammarGPT: Exploring Open-Source LLMs for Native Chinese Grammatical Error Correction with Supervised Fine-Tuning