LangChain’s Post

🦙 Tool calling with Ollama We now have a partner package with Ollama to help you perform tool calling, which is now natively supported in Ollama. Tools are utilities (like APIs or custom functions) that enhance an LLM's capabilities. However, local LLMs struggle with both selecting the right tool and providing the correct input. In the video below, we use the new Ollama partner package to perform tool calling w/ the recent Groq fine-tune of Llama-3 8b. See how to create a simple tool calling agent in LangGraph with web search and vector-store retrieval tools that run locally. 🎥 Video: https://lnkd.in/erppcmdY 🐍 Partner package (Python): https://lnkd.in/ej7KUQCr 🦏 Partner package (JavaScript): https://lnkd.in/eY8giWBY 📓 Notebook: https://lnkd.in/ewfNM_dc

This is the best of two - #ollama and #langchain framework.

LangChain bind_tools with Ollama models. Tool Calling one of the most widely used feature of LLMs for building real use cases on generative ai. Now, you can use this with Ollama local models. #ollama #toolcalling #ai

Tool Calling with Ollama: The Ultimate Power-Up! Tool calling is like having superpowers for your LLMs. It's like having a utility belt full of APIs, custom functions, and more! But, local LLMs can get overwhelmed with choosing the right tool and providing the correct input. That's where Ollama's partner package comes in! With this new package, you can create a simple tool calling agent in LangGraph that uses web search and vector-store retrieval tools that run locally. Check out our video below to see how it's done! Get ready to level up your LLM game with Ollama's tool calling partner package!

🚀 Project Highlight: Session Hijacking with DVWA 🚀 🔐 Objective: Demonstrate session hijacking to gain unauthorized login access using the Damn Vulnerable Web Application (DVWA). 📜 Description: Tools Used: DVWA, Burp Suite, Web Browser Steps: 1) Setup DVWA: Install and configure DVWA on a local server. 2) Login and Capture Session: Use Burp Suite to intercept and capture the session cookie during a login attempt. 3) Hijack Session: Replay the captured session cookie to gain unauthorized access. 🔧 Code Snippet: Python: # This is a conceptual representation of the steps involved # Actual implementation involves using Burp Suite and browser tools # Step 1: Capture the session cookie using Burp Suite session_cookie = "captured_session_cookie" # Step 2: Use the captured cookie to hijack the session headers = {   "Cookie": f"PHPSESSID={session_cookie}" } response = requests.get("http://dvwa.local/login.php", headers=headers) if "Welcome" in response.text:   print("Session Hijacking Successful!") else:   print("Session Hijacking Failed.") 🌟 Outcome: Successfully demonstrated session hijacking by gaining unauthorized access using a captured session cookie.

If you have a Python API 🐍 this might interest you! Identifying performance bottlenecks in your API can be challenging and I want to show you a good library: pyinstrument (https://lnkd.in/dQgXGXAn) With just a simple environment variable, you can set it up so that after each request, a detailed report opens in your browser, showing exactly how your application is performing and which functions are consuming the most. While there are several good profiling tools available, like PyCharm’s built-in profiler and cProfile, what stands out about pyinstrument is its versatility. It offers various ways to view your profiling data, whether in the terminal, through a web browser, etc. Plus, the user interface is intuitive and easy to navigate. (Here it is an example with FastAPI) Now, I’m curious: What profiling tools do you currently use for your API performance analysis?

This is a CLI calculator that I made using Inquirer with the help of this calculator you can easily add, subtract, multiply, divide, Exponent, or Modulus any values. Here you can check my git repo:[bit.ly/4bOJJWR]. You can also run this program by simply typing the command on cmd:-[npx bisma_shah_cli_calculator]. Sir Zia Khan | Sir Daniyal Nagori | Sir Ameen Alam | Sir Ali Jawwad | Sir Ahmed Raza Shaikh | Sir Hamzah Syed #Governor #Sindh #Initiative #for #GenAI, #Web3, #and #Metaverse #typescript, #javascript, #python, #react, #next, #SQL, #kafka, #dockers

Wouldn't it be great if you get an email when a product you are interested to buy gets discounted to a price you set? Using web scrapping which allows us to obtain data from websites, we can automate a Python code that gets run everyday by website like Pythonanywhere.com to send us an email if a product we are interested in gets discounted below a price that we set. You can try the code and modify it for a product of your choice and a price you set for it. Note: websites are built different so you will need to investigate your website for the price item in the website HTML. Code on Github: https://lnkd.in/dBK2RcCj

Escaping Config Hell In my experience when people need to configure a program, they start with some constants at the top of a file, and eventually end up with a configuration tool, a hell hole of json/toml, and custom serialization de-serialization logic. Though this is generally done in search of increased simplicity, usually it doesn't work out that way. There is a fundamental impedance mismatch between the config format and the language you are working in. A typo or misspelling can kill the config file. You will be provided zero sane errors, and types are not preserved. Not to mention your editor gives no hints at what the config file should look like. For this reason I sometimes like to configure a project within the native language a program is written. Mix this with a little bit of meta-programming, and it enables you to do some things that would otherwise be a compounding pain in the ass for the next few months / years. The attached image is an example of native configuration. This single decorator makes a clean pairing between files, their corresponding paths, and load function in very few characters. It also generates a data structure which caches the files automatically, which even supports pre-warming. #python #macros

I made a terminal based web browsing program, Terminal Search is a Python script designed to enhance web browsing efficiency by providing a seamless command-line interface for web searches and navigation. Leveraging the power of Python, BeautifulSoup, and fzf, Terminal Search offers a fast and intuitive way to search the web and browse selected results without leaving the terminal window. Key Features: Web Search: Conduct web searches using the Brave search engine directly from the command line. Interactive Selection: Utilize fzf for interactive selection of search results, enhancing user experience and efficiency. Seamless Navigation: Navigate to selected search results effortlessly within the terminal, opening them in the w3m text-based web browser. Customization: Easily configurable and extensible, allowing users to tailor the tool to their specific needs and preferences. Usage: Installation: https://lnkd.in/gci9Wsb2