there are a lot of catfish accounts on twitter which generally don’t bother me but this one person claiming to be a YC founder and an ex @/microsoft bothers me to my core because i also have a lot of mutuals (including CEOs and senior devs / designers) following her, please
deepsilicon (YC S24) is making hardware to run neural nets with orders of magnitude less latency, energy, and 5x less RAM. Right now, they're proving it with software.
Congrats on the launch,
@abhireddy2004
and
@nanda_alexander
!
How did Microsoft researchers tackle fine-tuning giant LMMs?
They created LoRA⚡
LoRA trains low rank weight matrices to reduce the # of trainable parameters (10,000x for GPT-3!) while maintaining performance.
Met a team of young founders that checked themselves into rehab.
They don’t even have an addiction problem they just wanted to grind:
- Live together
- No drugs/alcohol
- Healthy meals provided
- Scheduled group therapy sessions
I think
@klaviyo
is the most under-appreciated public company.
12 years ago, they couldn’t raise $20k. Today, they’re a SaaS giant with a $6.23BN market cap.
💰 39% YoY growth approaching $1BN in ARR.
🚀 $50,000 ACV customers are up 80%. Woah!
💸 Super capital efficient: 16%
So people *really* want to learn Cursor.
Already 1,056 people learning in my Cursor course.
And my “Building a pro full-stack app with AI” course launches this week to pair with it.
Come learn to build with AI.
25% launch discount - link below.
You *need* to try
@SupermavenAI
.
It's a huge improvement over Github Copilot.
INSTANT code suggestions.
Massive (300k) context length.
Watch me use it to build a Flask API in seconds.
There's a new SOTA model on the swe-agent leaderboard. Honeycomb just beat out Amazon Q with 22.06%. Devin's score was just 13.86%.
Here's who Honeycomb is and how they got sota.
Qwen-VL is an insane step up for Open Vision Language models
I tested Qwen-VL-Max on a doc -> JSON eval from the LLaVa 1.5 paper, and it beat out every other model including GPT-4V 🤯
Finally found an eval I like for LMMs:
Current leaderboard basically confirmed my qualitative eval that Gemini and GPT-4V are way ahead of any Open Source LMMs except maybe for CogVLM.
📣ANNOUNCING THE FASTEST AI CHIP ON EARTH📣
Cerebras proudly announces CS-3: the fastest AI accelerator in the world.
The CS-3 can train up to 24 trillion parameter models on a single device. The world has never seen AI at this scale.
CS-3 specs:
⚙ 46,225 mm2 silicon | 4
@twofifteenam
Tried building this for ArXiv papers once (kinda like ).
Realized just semantic search without the extras like author filters or keyword matches isn't that great tbh.
Introducing GPT-4o mini! It’s our most intelligent and affordable small model, available today in the API. GPT-4o mini is significantly smarter and cheaper than GPT-3.5 Turbo.
We are working to release Micro-Engine Scheduler(MES) documentation towards end of May and will follow up with published source code for external review and feedback. We have also opened a GitHub tracker, which will have the latest status on fixes and release dates.
Introducing Sora, our text-to-video model.
Sora can create videos of up to 60 seconds featuring highly detailed scenes, complex camera motion, and multiple characters with vibrant emotions.
Prompt: “Beautiful, snowy
🐍💥Python 3.13.0 has been released! 🎉 This is the first version with 🧵experimental GIL-free mode, an experimental JIT compiler🔧, a slick new REPL 🖌️ and many new cool features! And it's faster, smarter, and more colorful than ever! 🚀
Get it here:
// life update //
- girlfriend married someone else last month
- lost 15 kilos (sustaining 85 now)
- started MMA (boxing, muay thai soon)
- skill maxxing (started c, then py, soon design, hardware, economics, ++)
- actively working on projects/ideas
my life now revolves around