const currentPage = dv.current().file;
const dailyPages = dv.pages('"0-Daily"').sort(k=>k.file.name, "asc");
const currentPageName = currentPage.name;
const index = dailyPages.findIndex((e) => {return e.file.name === currentPageName});
if (index < 1) {
dv.table(["File", "Created", "Size"],[]);
} else {
const lastIndex = index - 1;
const lastPage = dailyPages[lastIndex].file;
const allPages = dv.pages().values;
const searchPages = [];
const lastTime = dv.parse(lastPage.name);
const currentTime = dv.parse(currentPage.name);
debugger;
for (let page of allPages) {
const pageFile = page.file;
if (pageFile.cday > lastTime && pageFile.cday <= currentTime) {
searchPages.push(pageFile);
}
}
dv.table(["File", "Created", "Size"], searchPages.sort((a, b) => a.ctime > b.ctime ? 1 : -1).map(b => [b.link, b.ctime, b.size]));
}
ExtractGPT
ExtractGPT
Chrome 扩展, 可以直接通过 GPT 自动提取网页内容。从而帮助解决网页限制
实际使用效果一般。猜测是通过对网页抽取后,匹配文字,然后进行总结得到的效果。
有时间可以看看实现原理
DeepSpeed-Chat
本章将演示如何在 DS-Chat 代码中使用除 facebook opt 之外的其他预训练模型,以及如何准备并使用自定义数据进行模型训练,以便训练出针对特定领域或应用的大型模型。
应该是一个教程,教学如何进行自定义训练的。
DeepSpeed-Chat 模型与数据 | 大模型训练入门实战
里面提到一个模型 Phoenix-inst-chat-7b , 和 GPT4 对该模型的评价
| Model | Ratio |
|---|---|
| Phoenix-inst-chat-7b vs. ChatGPT | 85.2% |
| Phoenix-inst-chat-7b vs. ChatGLM-6b | 94.6% |
| Phoenix-inst-chat-7b vs. Baidu-Wenxin | 96.8% |
| Phoenix-inst-chat-7b vs. MOSS-moon-003-sft | 109.7% |
| Phoenix-inst-chat-7b vs. BELLE-7b-2m | 122.7% |
| Phoenix-inst-chat-7b vs. Chinese-Alpaca-7b | 135.3% |
| Phoenix-inst-chat-7b vs. Chinese-Alpaca-13b | 125.2% |
更好的提示语
Useful verbs:
Instruct verbs are the first word in the prompt. They are used to instruct the model to do something. They are the most important part of the prompt. We may argue that they act as magic words that make a successful prompt, it helps to make the machine understand and follow your instructions. Each of these verbs has a specific meaning that can help the model understand what kind of answer is expected. For example, if you ask the model to explain something, it will provide a detailed explanation. On the other hand, if you ask the model to summarize something, it will provide a brief overview.
-
Useful words to give directions to your input texts: (point to a location of your input text)
\Given text
\following passage
\Following
\The following -
Useful words to instruct: (to give instructions to the model)
\Explain
\Create
\Generate
\Provide
\Write
\Summarize
\Paraphrase
\Translate
\Rewrite
\Reword
\Rephrase
\List
\Extract
\Pick out
\Analyze
\Define
\Identify
\Describe
\Predict
\Explain
\Analyze
\compare
\Categorize
\Evaluate
\Critique
\Differentiate
\Examine
\Illustrate
\Outline
\Calculate
\Infer
\Revise
\Summarize
\Classify
\Develop
\Evaluate
\Formulate
\Discuss
\Clarify
\Highlight
\Simplify
\Apply
\Suggest / Propose
\Make
* You may add `Can you` in the case of ChatGPT to make it more conversational, but it still works in both the instruct-model and chatGPT models. - Can you provide … - Can you generate …
Common use cases:
Classification
Instruction verbs that are useful for classification tasks in nlp
Classifythe following as a physical or chemical propertyIdentifymain idea of the given passageCategorizethe following passage as either a myth or a factAssign eachitem to its corresponding category based on its characteristicsListthe five main categories of the given data set.Groupthe items based on their common characteristicsSortthe data based on the date it was collectedLabeleach data point with its corresponding categoryDividethe data into smaller subgroups based on their similaritiesGroupthe items based on their similarity in featuresLabeleach group with a descriptive name.
Generation
Instruction verbs that are useful for generation tasks
Generatea new text based on the given parametersCreatea short catchy headline based on the following, aim to promote the productWritea story using the following keywordsFormulatea new sentence based on the given structureProducea new text that is similar in style to the given sampleConstructa new text that combines elements from multiple sourcesComposea new poem based on the given themeBuilda new text by combining different elements in a creative wayDevisea new narrative based on the given informationInventa new story using the following elements.
Transformation:
Instruction verbs that are useful for transformation tasks
- Translate:
Translatethe following function into C#Translatethe following into 1. English, 2. French, 3. Swedish
- Summarize:
Summarizethe main points of the given article
- Paraphrase:
Paraphrasethe following sentence to convey the same meaningRewordthe following sentence to make it more concise
- Simplify:
Simplifythe following equation for easier understanding
- Revise:
Revisethe following paragraph to make it more clear and concise
- Outline:
Outlinethe steps to solve the given problem
- Calculate:
Calculatethe total cost of the following items
- Infer:
Inferthe meaning of the [jargon, vage phrase] based on the following context:
- Apply:
Applythe following formula to calculate the resultApplygiven tone and voice to the passage below
Comparison:
Instruction verbs that are useful for comparison tasks
- Compare:
Compareand contrast the two given texts
- Contrast:
Contrastthe given passages to highlight their differences
- Differentiate:
Differentiatebetween the two given terms
- Examine:
Examinethe similarities and differences between the two given texts
- Analyze:
Analyzethe two given pieces and identify their similarities and differences
- Critique:
Critiquethe two given passages and compare their strengths and weaknesses
- Evaluate:
Evaluatethe two given passages and determine which one is more effective
Moment.JS
一个同时支持 js 和 node 的日期和时间库。上百万次使用。
Auto-Dev
AutoDev 是一款高度自动化的 AI 辅助编程工具。AutoDev 能够与您的需求管理系统(例如 Jira、Trello、Github Issue 等)直接对接。在 IDE 中,您只需简单点击,AutoDev 会根据您的需求自动为您生成代码。您所需做的,仅仅是对生成的代码进行质量检查。