I just shipped a project called Airfeed: https://airfeed.co/
Here's how it works:
1. You tell it the research topics you're interested in following
This can be anything -- LLMs, chatbots, image generation, scaling model inference, small language models, datasets, etc.
2. Every morning, it filters new AI research papers for quality work that is relevant to your configured topics, and sends you an email summary.
Here's a sample email: https://airfeed.co/example_email.html
---
As a fun experiment, I've made everything completely autonomous: Every morning, ChatGPT takes care of filtering/processing all research papers & writing each email.
---
Why I built this:
I think there needs to be a better way for engineers, researchers, and AI product builders to stay up to date on the latest AI research that's relevant to whatever they're working on.
For someone building LLMs, this could mean keeping up with research papers covering multiple topics, ex: scaling attention, model architecture, new datasets, new fine-tuning techniques, etc.
---
How this is different from other newsletters:
This isn't a generic AI newsletter.
It's research-focused & personalized, since it only highlights the papers that might be relevant to you (based on your specific topics).
I found this to be useful, since a lot of papers are irrelevant and/or low quality.
---
When building maroofy.com [1], I would regularly read the latest research papers on everything related to: contrastive learning, vector embeddings, music classification, etc.
So, I thought it would be nice to build something that filters/summarizes just the new research papers covering topics I care about, and sends me the top ones in a daily, morning email.
Would love any feedback!! :D
