Paul Graham’s essays have influenced the way I look into entrepreneurship.
I’ve condensed over 20 of his best into quick one-pagers for easy reference.
Starting with a classic: "Do Things That Don’t Scale."👇
Ilya Sutskever gave John Carmack a list of papers to read, so as to improve John's understanding of AI. That is a lot to read. If you want to quickly grasp what those papers are about, in this thread I share maps on over 20 of them 👇
Been following
@MarioNawfal
during the last hours to get a sense of what is happening in Rostov now. So many potential consequences so I decided to plot the most plausible ones with
@thevisualizerai
. we'll see how it evolves.
til, Ilya sutskever gave john carmack this reading list of approx 30 research papers and said, ‘If you really learn all of these, you’ll know 90% of what matters today.’
@IndieHackers
More complex and faster concept maps. GPT-4 was great for mindmaps (along Claude 3 and Llama 3). But it was not practical because of the lack of speed in returning JSON. But GPT-4o now flies :)
Been following
@MarioNawfal
during the last hours to get a sense of what is happening in Rostov now. So many potential consequences so I decided to plot the most plausible ones with
@thevisualizerai
. we'll see how it evolves.
🚨COUP UPDATES: Wagner with the National Guard & Russian Military TAKE OVER Russian Military HQ In Rostov
Tensions rise in Russia as the Southern Russian Military District Headquarters in Rostov is surrounded by armed military personnel.
The surrounding military personnel,
@paulg
@s8mb
It all starts with being obliged to have a representative from a Trade Union (over 50 employees). Same as in Spain.
Prompted it into my tool as this is the response:
@charliebholtz
@elevenlabsio
Had to try with one of the greatest, and also one of my child heroes, Dr. Félix Rodríguez de la Fuente. Result is mind blowing. Thanks for sharing, Charlie!
@RMcElhenney
@TheSunnyPodcast
Mind maps and knowledge graphs can be particularly beneficial in these cases, regardless of the of diagnosis. (Disclosure, the mind map below is generated by a tool I've built).
@HelenBevan
Insightful, Helen.
@MarkFritz
I prompted "create a comprehensive mind map of the impact of micromanagement on psychological safety" and here is the response:
@_lukaemon
@Kartikayb77
Yes it’s the highest quality thing, by a margin, I can find. Sometimes it goes a bit fast, exercise to the reader to fill in detail, otherwise very good.
This essay is charming and insightful, but it doesn't give any hint about how to *make* discoveries. For that, I would recommend this free PDF of a book by Professor Jack Oliver, one of the discoverers of plate tectonics.
Very interesting how this post is taking off. The one and only
@Scobleizer
shared it with his audience. That's it, as simple as that 😀. Thank you, Robert.
If any of you are interested in trying out The Visualizer, today I've launched a new 1-year access plan, for 29 euros. For a
Eye-opening talk between Vinay Prasad and Russ Roberts about "The Good, the Bad, and the Ugly of the Covid Vaccine". I've mapped out both the episode and the Q&A session:
Had to play with Robert's idea and works great. I Added text-to-speech to improve accessibility. I also wanted to make it 100% local so I used MacOS 'say'.
Considering my M2 with 8GB I'm pretty happy with the result. Local deployment a bit clunky, so time to check OpenAI
So I set up BakLLaVA-1 in the llama.cpp, and now it can provide real-time descriptions of the live feed from my camera
check it out!
open source?
cc:
@nisten
@thursdai_pod
@willdepue
#llama
Extra benefit of levaraging AI to grasp complex concepts quickly: inmediate translation. If you are not native in a language, this is incredibly helpful. Here is one of the papers (Attention Is All You Need) translated into French:
Ilya Sutskever gave John Carmack a list of papers to read, so as to improve John's understanding of AI. That is a lot to read. If you want to quickly grasp what those papers are about, in this thread I share maps on over 20 of them 👇
I was interested on Nik's approach to optimal landing pages so I decided to expand his 4 angles.
I've found many nodes where you can focus your attention.
Earlier today I roasted landing pages for 2 hours straight. The full video is below.
Most people had the same issues with their landing pages, PDPs, and homepages. Here are 4 improvements that would help 99% of people:
1. Use your product photos more wisely. Add text overlays
@NickADobos
Hi Nick, for my RAGs I am using Microsoft Azure Document Intelligence (very powerful OCR) and it works, always. No need to format documents. Using it with old legal docs; 100% accuracy with hybrid search (semantic + keyword).
@balajis
Indeed. Daniel Yergin's The Prize was eye-opening for me (I read it over 35 years ago). 100 years ago people thought oil was going to last for 75 years. And 75 years ago the same. We should never rule out human innovation 😀. (Disclosing my ideation tool: I prompted the Y-axis of
@simonw
Hi Simon, I have deployed this repo and works very well. Done it for a law firm with excellent results. I would invite you to got through the files in the picture.
@RishiSunak
Prompt: "I’m working hard to deliver on your priorities and build a better future for our children and grandchildren. Let’s run through another week of delivery"
Response:
@erhartford
I’m interested in this. It appears that Ekman’s theory of basic emotions (6 vs Plutchik’s 8) is more accurate. There is a lot of research comparing both.
Ekman later on proposed an expanded list of emotions. Interesting reading here:
The letter everybody is mentioning today worries me, not because of the "risks" LLMs pose to humanity -which are basically zero- but because it shows such an unhealthy cult inside OpenAI towards AGI. They've become religious...
A group of current, and former, OpenAI employees - some of them anonymous - along with Yoshua Bengio, Geoffrey Hinton, and Stuart Russell have released an open letter this morning entitled 'A Right to Warn about Advanced Artificial Intelligence'.
I don't get the hype with LLaMA2.
It's so focused on safety that it might end up being useless.
Makes me think being open-source might be related to its inherent bias. Could it be a subtle way to shape our perspectives?
Here's a comparison with ChatGPT. At least OpenAI allows
@bindureddy
Nicely put, Bindu. I've tried some approaches and a customised version of this repo from Azure OpenAI has shown awesome results. Great OCR from Azure AI Document Intelligence, it also retrieves the exact page of the response. Built with authentication, document level control
Introducing
It does only one thing and does it exceptionally well:
It transforms ChatGPT responses into mind maps. It identifes key topics and creates corresponding nodes, connected based on the relationships and context provided by the ChatGPT response
Huge upgrade today. The Visualizer maps audio files and YouTube videos.
Quick example with introductory video from Stanford CS 25 (Transformers United)
Big update for The Visualizer. New features 🚀
Map YouTube Videos and Podcasts in Seconds
Simply copy and paste a YouTube link into The Visualizer, and it will map out the content in seconds. Want to summarize a podcast or any audio file? Upload it to the tool, et voilà!
Save
Initial roadmap accomplished 🚀
The Visualizer (powered by GPT-4 32k) reads entire URLs. It extracts the most relevant concepts and relationships and then builds clear maps. Want to go deeper? Expand any node. Response is based on the content of the URL.
Of course, you can
As of today, Chat is also available in the Concept Map layout. You can now have conversations with your concepts maps and unveil connections you didn't know existed 🧠
How to go from 0 to $5M ARR profitably (step by step)
Here is every growth hack we used for each of our distribution channels:
- Organic Short-form content
- Influencer Marketing
- SEO
- Paid Ads
Organic Short-form content
The most important thing to remember in this new age
I don't get the hype with LLaMA2.
It's so focused on safety that it might end up being useless.
Makes me think being open-source might be related to its inherent bias. Could it be a subtle way to shape our perspectives?
Here's a comparison with ChatGPT. At least OpenAI allows
@VPrasadMDMPH
Thank you Vinay for sheding some light on the current state of relevant trials and health policies in general. Been following you for over a year and have learnt so much. A quick summary of the paper below:
@arankomatsuzaki
This is why Gemini is going up to 10M tokens context window? (internally, not publicly available yet). This is a very interesting paper, Aran, thanks for posting. Main takeaways 👇
Advancing High Resolution Vision-Language Models in Biomedicine
abs:
code:
Introduces a Llama3-Med model, that is trained with hierarchical image representations coming from a pretrained CLIP encoder, passed to the Llama3 model
@hpdailyrant
Hello Hà, this resonates to me. May I suggest a tool? Disclosure: I've built it. It is an AI concept and mind mapping tool. Very easy to synthesize large amounts of information. And you can plug in branches as you like (with natural language). Please see below a one-pager of your
Now you can map out YouTube videos with
Simply copy paste the link.
Here is a recent one from Andrew Huberman with Dr. Gary Steinberg: How to Improve Brain Health & Offset Neurodegeneration.
2 hours of video, 45k tokens, in 30 seconds.