My Python script 🤖 can write better commit messages than you 😉
Check out my blog post to see how to generate commit messages with GPT-3.
https://www.mikulskibartosz.name/generate-commit-messages-with-gpt-3/
Remote
Bartosz Mikulski
@mikulskibartosz@mathstodon.xyz
mastodon 4.7.2MLOps engineer * freelance DevRel for data/ML startups * blogger * conference speaker *
0 Followers
0 Following
16 Posts
Joined December 16, 2022
website:
Did you know that #GPT-3 can now access the internet? Yes, it's true! 🤯
No, I'm not talking about the #AI in Microsoft Bing. I'm stuck on their endless waitlist. 🙄 I needed something else. What is it?
I recently wrote an article demonstrating how to connect GPT-3 to the internet and use it to access websites and download their content. Using Slack and #Langchain, I created a Slack bot that could do a simple web search and even a short, multistep search.
https://www.mikulskibartosz.name/how-to-connect-gpt-3-to-internet/
Can you replace OpenAI GPT with an open-source model and get decent results? It turns out you can❗️
Check out my blog post about using the Cerebras model with LangChain.
We have seen people posting about the OpenAI Function Calling feature and showing us calls to a simple Python function. But can it do something useful?
What about an AI-powered Slack bot that calls a REST API?
Don’t let AI models keep making the same mistakes! Use Guardrails, an open-source Python package, to validate and correct the output of large language models. Get precise control over the output with custom validators and corrections!
https://www.mikulskibartosz.name/guardrails-ai-advanced-validators/
🧙How to extend the capability of ChatGPT and allow it to retrieve data from your applications? You can build a plugin! 🏗️
In my new article, I will show you how to build a ChatGPT plugin that searches for relevant information in a vector database.
👉 https://www.mikulskibartosz.name/build-chatgpt-plugin-in-python/
I have seen many people saying they will generate test cases using AI. That's the worst thing you can do! 🤯
However, using AI for test-driven development is quite a good idea.
In my new article, I will show you how to do it. 👇
Machine learning models learn only during training time, right? 🤖 Can they learn anything during inference? It turns out that models with attention, for example, #GPT, can learn at inference time.
Check out my new blog post to learn how to use in-context learning and how it works!
https://www.mikulskibartosz.name/in-context-learning-prompt-engineering/
Replace your data analyst with an AI Slack bot ;)
Or rather, stop bothering your data analyst with simple queries that even AI can answer.
If you want to build it for yourself, here is my tutorial: https://www.mikulskibartosz.name/ai-data-analyst-bot-for-slack-with-gpt-and-langchain/
Replying to
@apicultor@hachyderm.io @epixoip@infosec.exchange @alx@mastodon.mit.edu
Don't import those passwords. You have to change them anyway.
Also, when you change the password, it's good time to decide whether you still use the service. Maybe you can remove the accounts you no longer need
🤔 Struggling with extracting vital information from lengthy documents using AI? There's a catch: It's inefficient to pass them one by one, and often, they're just too long.
✂ Typical splitting methods mean that AI has access to only a chunk of the entire document at a time, which can destroy the overarching context.
Without the whole picture, the AI faces challenges grasping nuanced meanings, making accurate extraction an uphill task. It's like trying to understand a movie by only watching random scenes. 🎥
🧠 Dive into my latest article on "Finding information in long documents with AI using vector databases and MapReduceChain from Langchain". I explore how to seamlessly split content, maintain context, and harness Langchain to make your AI's task effortless.
You can write #AI
prompts as SQL. Yes, seriously.
LMQL is a Python implementation of a SQL-like language for AI prompts described in the “Prompting Is Programming: A Query Language for Large Language Models” research paper.
The library allows writing prompts for Large Language Models in the form of SQL queries. It even supports validation of the generated text using the where clause. SQL really works everywhere.
Take a look at my article about it!
https://www.mikulskibartosz.name/using-ai-in-python-with-lmql/
What does software architecture look like in AI-based systems? Do you need a guide to all components required while deploying AI-based systems?
In my new article, I described the components you need and explained why they are crucial.
https://www.mikulskibartosz.name/software-architecture-for-mlops-and-ai/
Do you have lots of "knowledge base" documents? What if you had an AI chatbot that answers questions using data from those documents?
Let me show you how to build a Facebook chatbot for basic customer support.
https://www.mikulskibartosz.name/build-ai-chatbot-for-facebook/
What do a text search and duplicate detection has in common❓You can use word embeddings and vector databases for both of them!
Check out my new article about using the Milvus vector database and OpenAI text embeddings API.