What is this?
A huge curated collection of Chinese NLP resources: corpora, lexicons, tools, datasets and courses. The README calls it 'The Most Powerful NLP-Weapon Arsenal' — nearly the most complete Chinese NLP resource library.
A huge curated collection of Chinese NLP resources: corpora, lexicons, tools, datasets and courses. The README calls it 'The Most Powerful NLP-Weapon Arsenal' — nearly the most complete Chinese NLP resource library.Is it a library I can pip install, or something else?It is a directory, not a package: the README is tables of resources, each entry giving a name, a description and a link to the resource's own repo — you follow links instead of installing anything here.
It is a directory, not a package: the README is tables of resources, each entry giving a name, a description and a link to the resource's own repo — you follow links instead of installing anything here.How do I find my way around something this big?It moves from ChatGPT/LLM topics (evaluation, frameworks, prompt engineering, safety, multimodal) to classic NLP sections: corpora, lexicons, extraction, knowledge graphs, summarization, QA, sentiment analysis and more.
It moves from ChatGPT/LLM topics (evaluation, frameworks, prompt engineering, safety, multimodal) to classic NLP sections: corpora, lexicons, extraction, knowledge graphs, summarization, QA, sentiment analysis and more.Is everything in the list still up to date?The author updates it irregularly over the long term and invites watches and forks; since every entry is a link to a third-party repo, freshness varies from resource to resource.
The author updates it irregularly over the long term and invites watches and forks; since every entry is a link to a third-party repo, freshness varies from resource to resource.Is this aimed at NLP beginners or at experienced practitioners?The author collected these packages while going from NLP beginner to practitioner and shared them for others on the same path, so both learners and working NLP engineers are the target crowd.
The author collected these packages while going from NLP beginner to practitioner and shared them for others on the same path, so both learners and working NLP engineers are the target crowd.Which LLM benchmarks does it collect?The evaluation section lists C-Eval, a Chinese suite of 13,948 multiple-choice questions across 52 subjects at four difficulty levels, plus OpenCompass from the Shanghai AI Laboratory and Chatbot Arena's Elo-rating leaderboard.
The evaluation section lists C-Eval, a Chinese suite of 13,948 multiple-choice questions across 52 subjects at four difficulty levels, plus OpenCompass from the Shanghai AI Laboratory and Chatbot Arena's Elo-rating leaderboard.Are there runnable ChatGPT-style frameworks in it?Yes: LLM-As-Chatbot runs a range of LLMs as chatbots directly on Google Colab with no setup, OpenBuddy is an open multilingual chatbot model in 7B and 13B sizes, and LaMini-LM offers small models distilled from ChatGPT on a 2.58M-instruction dataset.
Yes: LLM-As-Chatbot runs a range of LLMs as chatbots directly on Google Colab with no setup, OpenBuddy is an open multilingual chatbot model in 7B and 13B sizes, and LaMini-LM offers small models distilled from ChatGPT on a 2.58M-instruction dataset.What classic Chinese resources are inside?The repo description lists sensitive-word and stopword lists, synonym, antonym and sentiment lexicons, Chinese word vectors, an ancient-poetry corpus, a Xinhua dictionary database with an API, and tools that extract phone numbers, IDs and emails.
The repo description lists sensitive-word and stopword lists, synonym, antonym and sentiment lexicons, Chinese word vectors, an ancient-poetry corpus, a Xinhua dictionary database with an API, and tools that extract phone numbers, IDs and emails.Is it useful if my project is in English, not Chinese?Chinese is the focus, but English material is mixed in: Chinese-English sensitive-word lists, Chinese-Japanese-English name lexicons, the cs224n deep-learning NLP course, and word2word pairs covering 62 languages.
Chinese is the focus, but English material is mixed in: Chinese-English sensitive-word lists, Chinese-Japanese-English name lexicons, the cs224n deep-learning NLP course, and word2word pairs covering 62 languages.What are similar projects?NLP-progress, awesome-nlp, Chinese-Word-Vectors, HanLP, lectures, PaddleNLP, flair, bert
NLP-progress, awesome-nlp, Chinese-Word-Vectors, HanLP, lectures, PaddleNLP, flair, bert