NLP Course Reviews for Simplified Chinese (zh-CN)

[chenglu]

Hi there 👋

Let's review the zh-CN translation of the course, so that the whole community can benefit from this resource 🌎!

As well as fix: #118 ;-D

Below are the chapters and files that need reviewing - let us know here if you'd like to review any and we'll add your name to the list. Once you're finished, open a pull request and tag this issue by including #552 in the description.

💯 Considering that this is a course designed for beginners, it might be helpful to provide necessary explanations for some technical terms used throughout the course, making the content more accessible to the target audience.

The review work for this NLP course is also considered one of the "onboarding tasks" for new translation volunteers.

The recommended PR title is: docs(zh-cn): [NLP Course] Reviewed {{File_Name}}

like:

docs(zh-cn): [NLP Course] Reviewed `chapter0/1.mdx`

Note:

1. Please claim one course review item at a time.
2. Considering it will take times to review your PR, please try to 1) Fork this repo 2) Start a new branch to review 3) Open a Pull Request from that branch.
3. Let the team know if you have any questions here on the Discord chinese-simplified-translation channel.

Chapters

0 - Setup

• chapter0/1.mdx - 课程简介

1 - Transformer models

• chapter1/1.mdx - 本章简介
• chapter1/2.mdx - 自然语言处理
• chapter1/3.mdx - Transformers 能做什么
• chapter1/4.mdx - Transformers 是如何工作的
• chapter1/5.mdx - “编码器”模型
• chapter1/6.mdx - “解码器”模型
• chapter1/7.mdx - 序列到序列模型
• chapter1/8.mdx - 偏见和局限性
• chapter1/9.mdx - 总结
• chapter1/10.mdx - 章末小测试

2 - Using 🤗 Transformers - WHOLE chapter claimed by @wwzhuang01

• chapter2/1.mdx - 本章简介
• chapter2/2.mdx - 管道的内部
• chapter2/3.mdx - 模型
• chapter2/4.mdx - 标记器
• chapter2/5.mdx - 处理多个序列
• chapter2/6.mdx - 把它们放在一起
• chapter2/7.mdx - 基本用法完成
• chapter2/8.mdx - 章末小测试

3 - Fine-tuning a pretrained model - WHOLE chapter claimed by @Decem-Y

• chapter3/1.mdx - 本章简介
• chapter3/2.mdx - 处理数据
• chapter3/3.mdx - 使用 Trainer API 微调模型
• chapter3/3_tf.mdx - 使用 Keras 微调一个模型
• chapter3/4.mdx - 一个完整的训练
• chapter3/5.mdx - 微调，检查
• chapter3/6.mdx - 章末小测验

4 - Sharing models and tokenizers

• chapter4/1.mdx - The Hugging Face Hub
• chapter4/2.mdx - 使用预训练模型
• chapter4/3.mdx - 共享预训练模型
• chapter4/4.mdx - 构建模型卡片
• chapter4/5.mdx - Part 1 完结
• chapter4/6.mdx - 章末小测试

5 - The 🤗 Datasets library

• chapter5/1.mdx - 本章简介
• chapter5/2.mdx - 如果我的数据集不在 Hub 上怎么办
• chapter5/3.mdx - 是时候来学一下切片了
• chapter5/4.mdx - 大数据? 🤗 Datasets 来救援
• chapter5/5.mdx - 创建自己的数据集
• chapter5/6.mdx - 使用 FAISS 进行语义搜索
• chapter5/7.mdx - 🤗 Datasets 回顾
• chapter5/8.mdx - 章末小测试

6 - The 🤗 Tokenizers library

• chapter6/1.mdx - 本章简介
• chapter6/2.mdx - 根据已有的tokenizer训练新的tokenizer
• chapter6/3.mdx - 快速标记器的特殊能力
• chapter6/3b.mdx - QA 管道中的快速标记器
• chapter6/4.mdx - 标准化和预标记化
• chapter6/5.mdx - 字节对编码标记化
• chapter6/6.mdx - WordPiece 标记化
• chapter6/7.mdx - Unigram标记化
• chapter6/8.mdx - 逐块地构建标记器
• chapter6/9.mdx - 标记器回顾
• chapter6/10.mdx - 章末小测试

7 - Main NLP tasks

• chapter7/1.mdx - 本章简介
• chapter7/2.mdx - Token 分类
• chapter7/3.mdx - 微调掩码语言模型
• chapter7/4.mdx - 翻译
• chapter7/5.mdx - 提取文本摘要
• chapter7/6.mdx - 从头开始训练因果语言模型
• chapter7/7.mdx - 问答
• chapter7/8.mdx - 精通自然语言处理
• chapter7/9.mdx - 章末小测试

8 - How to ask for help

• chapter8/1.mdx - 本章简介
• chapter8/2.mdx - 出现错误时该怎么办
• chapter8/3.mdx - 在论坛上寻求帮助
• chapter8/4.mdx - 调试训练管道
• chapter8/4_tf.mdx - Debugging the training pipeline
• chapter8/5.mdx - 如何写一个好问题
• chapter8/6.mdx - 第2部分完成
• chapter8/7.mdx - 章末小测试

Events

• events/1.mdx - Live sessions and workshops (need translation)
• events/2.mdx - Part 2 发布活动
• events/3.mdx - Gradio Blocks Party (need translation)

[Decem-Y]

I chose chapter 3 for translation calibration 🤗.

[wwzhuang01]

I chose chapter 2 for translation calibration 🤗.

[Chaochao2020]

Subject: Feedback on Chinese Translation of the Hugging Face NLP Course

Dear Hugging Face Team,

I hope this email finds you well. I am writing to provide feedback on the Chinese translation of the NLP Course available on your website. As an enthusiastic learner, I was excited to explore the course content in Chinese. However, I noticed that the translation appears to have been done using machine translation, resulting in several issues that make it challenging for readers like me to learn effectively.

Firstly, there are clear traces of machine translation throughout the course. The phrasing often feels unnatural and does not flow smoothly, which can be confusing and distracting. Additionally, some sentences suffer from grammatical errors, making it difficult to grasp the intended meaning.

Furthermore, certain technical terms and concepts have not been accurately translated, leading to potential misunderstandings for learners. As a result, it becomes challenging to follow the course material and grasp the core concepts effectively.

I understand that translation can be a complex task, and I want to acknowledge the effort put into making the course available in multiple languages. However, I believe that addressing these issues will significantly enhance the learning experience for Chinese-speaking students.

I would like to offer some constructive suggestions to improve the translation quality:

Consider using professional human translators who are proficient in both NLP and the target language. This will help ensure accurate and contextually appropriate translations.

Conduct thorough proofreading and editing to eliminate any grammar or phrasing errors that might have been introduced during the translation process.

Establish a review process involving native speakers of the target language to provide valuable feedback and ensure the highest quality of translation.

If possible, provide an option for users to switch to the original English version if they find the Chinese translation too challenging to comprehend.

I want to emphasize that despite the translation issues, I appreciate the wealth of knowledge and resources available on your platform. I believe that with improvements to the Chinese translation, more learners will benefit from your exceptional NLP Course.

Thank you for considering my feedback. I am hopeful that you will take my suggestions into account to enhance the Chinese version of the NLP Course. If there are any updates or changes made, I would be delighted to revisit the course and recommend it to others in my community.

Once again, thank you for your dedication to providing valuable educational content. I look forward to witnessing the growth and improvement of the Chinese translation.

Best regards,
[chaochao]
[[email protected]]

主题：关于 Hugging Face NLP Course 中文翻译的反馈

亲爱的 Hugging Face 团队，

你好。我写信是为了就贵网站上 NLP Course 的中文翻译提供一些反馈。非常冒昧地在这个 issues 上给出这段话, 这是因为我仅仅通过 Github 上 找到了这里, 但没有找到您的邮件(只有你的推特账号). 作为一个积极学习者，我对用中文探索这门课程内容充满期待。但是，我发现翻译似乎带有机器翻译的痕迹，导致了一些问题，让像我这样的读者学习起来有些困难。

首先，课程中明显能看到机器翻译的痕迹。句子表达常常不够自然，缺乏流畅性，这可能会令人感到困惑和分心。此外，一些句子存在语法错误，让我们难以理解其真实含义。

其次，某些技术术语和概念并没有得到准确的翻译，导致学习者可能会对此产生误解。这让我们很难跟上课程内容，准确掌握核心概念。例如在第 9 章的 Gradio 的章节中很多专业术语都翻译的不准确, 一些细节也很不好, 比如说 英文和中文之间没有空格, 还有 中文版本没有相应的 colab 链接, 很多脚本并没有展示, 仅仅是 "Your space is on error, check its status on hf.co" 这句话等.

最后, 仅仅是 NLP Course 有相应的中文版本, 至于其他的课程以及文档都没有中文版本, 我觉得这并不是一个好的事情.

我理解翻译是个复杂的任务，同时我也想感谢您为了让更多人能接触到课程而进行多语言翻译的努力。然而，我相信解决这些问题将大大提升中文学习者的学习体验。

我想提供一些建设性的建议，以改进翻译质量：

1. 考虑使用专业的人工翻译人员，他们精通 NLP 和目标语言，以确保准确和符合上下文的翻译。

2. 进行彻底的校对和编辑工作，消除可能在翻译过程中引入的语法或表达错误。

3. 设立一个由目标语言的母语使用者组成的审核流程，以提供宝贵的反馈，确保翻译质量达到最高水平。

4. 如果可能，为用户提供切换到原版英文版本的选项，以防中文翻译过于困难理解。

我想强调，尽管存在翻译问题，我依然感谢您的平台提供了丰富的知识和资源。我相信通过改进中文翻译，更多的学习者将从您出色的 NLP Course 中受益。作为一个母语是非英文的人, 虽然我有阅读英文的水平(虽然水平并不高), 但我更倾向于中文的阅读和学习, 毕竟这是我的母语.

感谢您考虑我的反馈。我希望您能采纳我的建议，改进 NLP Course 的中文翻译。如果有任何更新或更改，我将很高兴重新学习这门课程，并向我所在的社区推荐。

再次感谢您致力于提供有价值的教育内容。我期待着看到中文翻译的进步和改善。

最诚挚的问候，
【超超】
【[email protected]】

[yaoqih]

Dear Chaochao,

Hello, I am one of the initiator of the Huggingface Course translation project. I'm glad to be able to provide feedback on the course, point out some current issues, and offer some targeted suggestions. Last year, we came across this course and benefited from it. We wanted to contribute to the community by helping beginners, just as we benefited from excellent Chinese translations when we started learning computers. However, due to time conflicts, learning pressure, and personnel changes, we didn't complete the translation of the second half, resulting in machine translation in that part.

The good news is that we reorganized the translation in June this year. This time, we are retranslating and proofreading all the content of the course to improve the quality and accuracy of the translation. So far, the work is nearing completion, and if everything goes well, a brand-new version should be available in mid-August.

In this version, we will address the following issues:

A translation team composed of native Chinese speakers with experience in the NLP field will review the content sentence by sentence, correcting mistranslations and inappropriate expressions.
We will refer to other translation organizations such as Jiqizhixin and establish our own terminology list to ensure the accuracy and universality of key proprietary term translations.
We will establish a review mechanism, with each piece of content being reviewed at least twice to ensure accuracy.
We will update outdated and fix broken links in the text.
Once again, thank you for your attention and support for the Huggingface Course. Translation is a complex task that requires continuous investment and optimization. Please stay tuned for the release of the brand-new version of the course in mid-August. If you encounter any issues while reading, please feel free to raise them as issues, and we will continue to pay attention and actively improve. We look forward to providing excellent course resources for the community together.

Best regards,
[Qi]

亲爱的超超，
你好，我是Huggingface Course翻译的发起人之一。很高兴能够对课程提供反馈，指出目前存在的一些问题并且提供一些针对性的建议。在去年我们接触到了这个Course，并从其中受益；我们想为社区能够贡献自己的力量以帮助初学者，就像我们当年入门计算机的时候也从大量的优秀的中文翻译中受益。但是在后续的时间，因为时间冲突、学习压力和人员变动，我们没有完成后半部分的翻译，因此后半部分的出现了机翻。
一个好的消息是我们在今年的6月份重新组织了翻译，这次我们将对Course所有的内容进行重新的翻译和校对，以提高翻译的质量和准确度。到目前为止，工作已经接近了尾声，顺利的话应该可以在八月中旬看到一个全新的版本。
在这个版本中我们将对以下的问题进行处理：

1. 由中文母语并在NLP领域有一定经验组成的翻译团队，对内容进行逐句地校审，修正错误翻译，不当的表达。
2. 参考其他的翻译组织如机器之心，建立自身的术语表，确保关键专有名词翻译的准确性和通用性。
3. 建立审核机制，所有内容至少完成2次校审，以确保内容的准确性。
4. 更新过时的文中链接，修正失效的链接。
   再次感谢你对Huggingface Course 的关注和支持，翻译是一个复杂的任务，需要持续的投入和优化。欢迎在本月中旬关注Course全新版本的发布，如果在那时阅读中遇见任何问题，欢迎再次提出issue，我们将持续关注并积极改进。期待我们可以一起为社区提供优秀的课程资源。

最诚挚的问候，
【淇】