Some surprising facts about new OpenAI Whisper Large V2 👇🏾

1. Whisper Large v2 makes 55.2% less errors on average than its Wav2Vec 2.0 counterpart

For some datasets like Common Voice you can see, Whisper Large v2 makes almost 70% lesser error than Wav2Vec2 (from Meta) 2. Whisper Large v2 follows the same architecture *just* extra epochs with regularization. That means these models would further better with new fine-tuning. If you're interested @huggingface's got a Whisper fine-tuning Sprint discuss.huggingface.co/t/open-to-the-…

I tried training a "Peppa Pig" #stablediffusion model using Dreambooth Unfortunately it's not good, but it's still fun.

Some Photos: 👇🏾

1. Peppa Pig + Tesla Car 2. Peppa Pig in front of Taj Mahal

Stable Diffusion Inpainting Tutorial using the new @runwayml model with a @Gradio App!

Full Tutorial -

Seems better than before! Edited out the Inference time which is about ~13 seconds on Colab T4

In 2022, Here's the easiest way to build your AI Product MVP in less than 24 hours. I'm going to show you how you can do with the 🤗 @huggingface ecosystem - all for Free to start! 👇🏾🧵 What do you need for an AI Startup MVP (Minimum Viable Product) ?
💡. Idea
🤖. Solution aka ML Model
💻. GUI
⚙️. App Hosting
🚀. Option to Scale

An MVP is one of the best ways to validate a product idea and in fact if you want to gain interest from users or Investors.

A thread of Free courses (that I believe are of good quality and $0 - atm)

CS50: #ComputerScience Courses and Programs from Harvard - edx.org/cs50 - @edXOnline Automate the Boring Stuff with Python - automatetheboringstuff.com - @AlSweigart