liltom-eth / llama2-webui

Run any Llama 2 locally with gradio UI on GPU or CPU from anywhere (Linux/Windows/Mac). Use `llama2-wrapper` as your local llama2 backend for Generative Agents/Apps.
MIT License
1.96k stars 202 forks source link

Test log | Welcome to communicate #9

Open XksA-me opened 1 year ago

XksA-me commented 1 year ago

test log

Cloud platform: matpool.com Machine used: NVIDIA A40 Model used: Llama-2-13b-chat-hf After the model is loaded, it takes up video memory: about 26G Inference memory usage: about 26G GPU utilization: about 80% image Memory usage: about 2G image

If hhhh cannot be written, it will also be filled with characters image

liltom-eth commented 1 year ago

Thanks for the test! Do you mind explaining more about what happened in the chatbot?

liltom-eth commented 1 year ago

Would you mind posting some English chat logs here? Since the original Llama2 is not supporting Chinese well I think.

XksA-me commented 1 year ago

Sure~

what can i do with you
Help me write a python crawler article
For a person with Python programming ability, how to learn go language?
In a cage, chickens and rabbits add up to 35 heads and 94 feet. Please count the number of chickens and rabbits.

image

Q: Help me write a python crawler article


A: Sure, I'd be happy to help! Here's an outline for a Python crawler article that covers the basics of building a web crawler using Python:

I. Introduction

Briefly introduce the topic of web crawling and why it's useful Explain what a web crawler is and how it works Introduce Python as a popular language for building web crawlers II. Why Python is a great choice for web crawling

Discuss the advantages of using Python for web crawling, such as its ease of use, flexibility, and extensive libraries Explain how Python's built-in modules, such as requests and BeautifulSoup, make it easy to handle HTTP requests and parse HTML and XML documents III. Setting up your development environment

Explain how to install Python and the necessary libraries on your computer Provide tips for setting up your development environment, such as using a code editor or IDE IV. Understanding the basics of web crawling

Explain how web pages are structured and how they are linked together Discuss the different types of web pages and how they can be crawled, such as HTML, CSS, and JavaScript V. Building a simple web crawler

Show how to use Python's requests module to send HTTP requests to a web page Explain how to use BeautifulSoup to parse the HTML content of the web page and extract the desired information Provide an example of a simple web crawler that crawls a list of URLs and extracts the title of each web page VI. Handling links and following them

Explain how to use Python to navigate the links on a web page and follow them to other web pages Discuss the challenges of handling links, such as dealing with redirects and broken links Provide an example of a web crawler that follows links and crawls a list of web pages VII. Crawling with user authentication

Explain how to use Python to login to a web page and crawl content that is only available to authenticated users Discuss the challenges of crawling with user authentication, such as dealing with CAPTCHAs and rate limiting Provide an example of a web crawler that logs in to a web page and crawls content that is only available to authenticated users VIII. Handling JavaScript and other dynamic content

Explain how JavaScript and other dynamic content can make web crawling more challenging Discuss the techniques for handling JavaScript and other dynamic content, such as using Selenium and Puppeteer Provide an example of a web crawler that uses Selenium to handle JavaScript and crawl a web page IX. Conclusion

Summarize the main points of the article Discuss the importance of web crawling and the potential applications of web crawlers Encourage readers to try building their own web crawlers using Python. X. Additional Resources

Provide links to additional resources, such as tutorials and documentation, for readers who want to learn more about web crawling and Python. This outline should provide a good structure for your article on building a Python crawler. Be sure to include examples and explanations to make the concepts more concrete and easier to understand. Good luck with your article!

> Q: For a person with Python programming ability, how to learn go language?
```bash
As a Python programmer, learning Go (also known as Golang) can be a rewarding experience. Go is a statically typed, compiled language developed by Google that is known for its simplicity, reliability, and performance. Here are some steps to help you get started with learning Go:

Familiarize yourself with the basics of Go:
* Start by reading the official Go documentation, especially the "A Tour of Go" tutorial.
* Understand the basic syntax and data types in Go, such as variables, constants, arrays, slices, maps, and structs.
* Learn about the built-in functions and control flow statements in Go, such as if/else statements, for loops, and goroutines.
Get comfortable with the Go environment:
* Download and install the Go programming language on your computer.
* Learn how to write, build, and run Go programs using the go command line tool.
* Understand how to use the gofmt tool to format your Go code.
Learn about Go's type system:
* Go has a statically typed type system, which means that you need to declare the types of your variables before using them.
* Learn about the basic types in Go, such as int, string, and bool, and how to use them.
* Understand how to use interfaces and type aliases to define custom types.
Learn about Go's concurrency model:
* Go has built-in support for concurrency, which allows you to write programs that can run multiple tasks simultaneously.
* Learn about goroutines and channels, which are the fundamental building blocks of Go's concurrency model.
* Understand how to use the sync package to synchronize access to shared data.
Learn about Go's standard library:
* The Go standard library provides many useful functions and types that you can use in your programs.
* Learn about the main packages in the standard library, such as math, strings, and os.
* Understand how to use the library to perform common tasks, such as reading and writing files, networking, and database access.
Practice writing Go programs:
* Start by writing simple programs to get a feel for the language.
* Gradually work your way up to more complex programs, using the concepts and techniques you've learned.
* Practice writing programs that use concurrency, input/output, and the standard library.
Join a Go community:
* There are many online communities and resources available for learning Go.
* Join online forums, such as Reddit's r/golang community, to ask questions and get help from other Go programmers.
* Participate in online meetups and events to learn from other developers and share your own knowledge.
Read Go documentation and resources:
* The official Go documentation is a great resource for learning the language.
* Read the Go blog to stay up-to-date with the latest developments and best practices in the Go community.
* Check out other resources, such as online courses, books, and tutorials, to further your learning.
Work on projects:
* One of the best ways to learn Go is to work on real-world projects.
* Start with small projects, such as command-line tools or web servers, and gradually work your way up to more complex projects.
* Use your projects as an opportunity to practice what you've learned and to apply new concepts and techniques.
By following these steps, you can gain a solid understanding of the Go programming language and become proficient in using it for a wide range of applications. Good luck!

Q: In a cage, chickens and rabbits add up to 35 heads and 94 feet. Please count the number of chickens and rabbits.

image

XksA-me commented 1 year ago

The answer to the last question is too funny. I don't know why it's like this. If you have anything else you want to ask, you can reply and I'll ask.

In addition, I would like to ask, how to set up single-machine multi-card or multi-machine multi-card operation.

XksA-me commented 1 year ago

I tested Llama-2-7b-chat-hf again today. in #17

liltom-eth commented 1 year ago

@XksA-me welcome contributing your benchmark performance here.