ChatGPT productivity, and then advanced! Now, throw a PDF in the past, it can read itself , and you are only responsible for asking questions.
A paper full of professional vocabulary can be explained clearly in a few words, and it is all in plain Chinese.
And the accuracy is amazing!
In the paper, LLaMA has several sizes, and the performance comparison results with GPT-3, it is not wrong.
Not just papers, such as contracts, documents, books, etc., as long as they are PDFs, they can support 200-page files.
This is the latest chatPDF , which analyzed 65,000 PDFs in just 5 days.
No need to register and log in now, it’s free for everyone !
It has to be said that the ecological growth rate of ChatGPT is really fast.
Interpret professional papers in seconds
The operation of ChatPDF is very simple, upload your own PDF to the page, it will start loading analysis, and then switch to the question interface.
As soon as it comes up, it will first “introduce itself”. And briefly interpret the content in the PDF, and list the key points one by one.
What we are uploading is a biological paper published in Nature, the main research is why mosquitoes like to bite people.
READ ALSO: Mi 6 mobile phone porting new progress of open source Hongmeng OpenHarmony
Then, we asked ChatPDF to explain the content of the paper in Chinese.
In just a few words, it clarifies the core points of the paper, from the basic principles to the experimental methods and the results of the paper.
And in just a few seconds, the answer is given.
You know, this paper has 23 pages including supplementary materials, and it will take nearly half an hour to figure it out.
If you just want to use it as a PDF built-in intelligent search engine is also OK.
For example, if you ask it who is the author of this paper, it can answer the name and point out the specific location in the paper.
(I got it right)
Or as a smart translator?
After experiencing it, the whole process is very silky~
According to reports, the principle of ChatPDF is to first analyze the uploaded PDF and create a semantic index for each paragraph in the file.
When the user asks a question, the tool will send the associated segment to ChatGPT, and then let it interpret it in conjunction with the question.
Like ChatGPT, the content of the interpreted PDF can be saved in the cloud for a period of 7 days .
The tool calls the author’s own ChatGPT API.
So there is also a big V appeal, everyone save some money… don’t waste tokens (banknotes).
However, this tool has its imperfections.
Some people said that it would cause incomplete reading of the article (and another tool was successfully used).
Moreover, it does not have the ability to read pictures . If the paper begins with a picture, it may lead to garbled interpretation.
And the old problem of ChatGPT-nonsense, still can’t be completely avoided.
You need to check it yourself when you use it.
Currently, ChatPDF is still being updated continuously. It originally had a limit of 50 pages for reading PDFs, but now it has been extended to 200 pages.
The developer is a German brother Mathis Lichtenberger . He has also developed other data processing tools firefoo before , and he claims that many big companies are using it.
One More Thing
It is worth mentioning that Google recently developed a similar tool called Flan5 LLM .
Based on the LLM model, it can be used on Colab.
How to put it, it seems that “I don’t want to read PDF word by word” is really the common voice of human beings.
There are also netizens who say that they have rubbed their hands with similar tools, so don’t be too cool~
Experienced children’s shoes are welcome to share your feelings in the comment area~
Portal:
ChatPDF : https://www.chatpdf.com/?continueFlag=cc734774b142837d5cb04bc1df00296d
Flan5_LLM: https://colab.research.google.com/drive/1AVh9dOsG9DKzfK7gOFrJuitPIcLPqlbO?usp=sharing
Source – IT Home
