Hey guys, Script Fury is up on Product Hunt today! 🎉
We would be stoked if you could give us an upvote and hit us up with some questions. Show us some love and help us crush it! 💪
#AI#NLP#gpt3#ProductHunt
https://t.co/m9K4qUSGBO
We are reproducing Flamingo, a vision and language model developed by Deepmind (https://t.co/GeLI64VN71).
We spent a good amount of time fighting training divergences (aka "instabilities"). Surprisingly, even at the ~2-3B scale.
Some learnings from overcoming these 🧵:
ChatGPT (gpt-3.5-turbo) works best for conversational tasks, InstructGPT (text-davinci-003) for zero-shot, and Codex (Code-DaVinci-002) for few-shot in-context learning.
As an NLP researcher, I am still valuable today, at least as a three-way classifier.
If you spend much time on AI twitter, you might have seen this tentacle monster hanging around. But what is it, and what does it have to do with ChatGPT?
It's kind of a long story. But it's worth it! It even ends with cake 🍰
THREAD:
We are excited to announce our new AI-powered scriptwriting tool that can help screenwriters to generate scripts for movies, TV shows and commercials. If you are interested, you can check out our website and join the wait list.
#AI#AItools#scriptwriting
https://t.co/9fGfuXvkOT
Deep Learning with PyTorch - University of Amsterdam(UvA)
A fantastic series of tutorials covering a wide array of topics from PyTorch basics, basics of neural nets, architectures(CNNs, transformers, GNNs), generative networks, and contrastive learning.
https://t.co/d6E9K1NYwC
I keep finding new uses for https://t.co/s9efw6geFx’s “Bird SQL”.
After a tweet of yours blows up, run the query: “Top followers of <your name> who are not followed back.”
Gives you a priority list after the storm of cool people you missed.
https://t.co/p5SFaCQN0Q
📣 New paper — In ‘Unnatural Instructions’ we propose a new method to automatically generate natural language instructions, allowing us to scale up to 240K diverse instructions & train models that rival the performance of contemporary instruction-tuned models.
📝 Read on ArXiv
What the crypto to AI migration shows us is just *how* many people are just excited about excitement.
“Everyone’s excited about ____. Seems really interesting. I’m starting to spend a bunch of time there!”
Tourists everywhere!
Announcing data2vec 2.0, a new general self-supervised algorithm built by Meta AI for speech, vision & text that can train models 16x faster than the most popular existing algorithm for images while achieving the same accuracy.
Read more & get the open source code ⬇️
GPT-4 is rumored to be coming soon, sometime between Dec - Feb
- GPT-3 has 175 billion parameters
- GPT-4 supposedly has 100 trillion parameters
It is something like 500x more powerful than GPT-3
What kinda stuff will you be able to create with GPT-4!?