Trying to achieve 100% efficiency is a bad thing?
There's always trade offs... can't violate the laws of physics
I am comfortable with 50% efficiency when building and always optimize afterwards.
I am starting to treat the computer like it completes tasks in light speeds... which it does lol
Past me had compute OCD... and it really held me back from getting stuff done sooner
Its perfectly fine and not a sin to be inefficient with compute... instead of forcing it just let the efficient way of doing things flow through you... if not, dw you'll optimize later. Its more fun that way
There's always trade offs... can't violate the laws of physics
I am comfortable with 50% efficiency when building and always optimize afterwards.
I am starting to treat the computer like it completes tasks in light speeds... which it does lol
Past me had compute OCD... and it really held me back from getting stuff done sooner
Its perfectly fine and not a sin to be inefficient with compute... instead of forcing it just let the efficient way of doing things flow through you... if not, dw you'll optimize later. Its more fun that way
the urge to build a notion clone desktop app with tauri sveltekit frontend and convex backend 🫠
⚡3
Dagmawi Babi
Ask ChatGPT "Is there a seahorse emoji?" and watch it go wild.
Here's my chat
Here's my chat
Call me bat shit crazy but this actually looks like signs of real intelligence
Like it's tweaking into self awareness 😂
Like it's tweaking into self awareness 😂
💘1
I would totally buy a smartphone that only has LLMs on it and nothing else... A good screen and SOTA models...
Dream Gadget right here
> You get a thought or idea
> Pull it out
> just start typing
Also personalized models are the future. High agency people would pay really good money to have a model trained on their data just like Mathew Mcconaughey said. This could be the interface.
there's no need to even touch the screen... you're just typing 90% of the time
> You get a thought or idea
> Pull it out
> just start typing
Also personalized models are the future. High agency people would pay really good money to have a model trained on their data just like Mathew Mcconaughey said. This could be the interface.
there's no need to even touch the screen... you're just typing 90% of the time
I just discovered comfy on reddit
It does everything that I am building but with a focus on open source diffusion models
Their barrier to entry looks much higher yet they have 90K stars on GitHub
It does everything that I am building but with a focus on open source diffusion models
Their barrier to entry looks much higher yet they have 90K stars on GitHub
www.comfy.org
ComfyUI | Generate video, images, 3D, audio with AI
The most powerful and modular visual AI application and engine
🔥4
will to code
Hot take AI + Image will be bigger than AI + Code Who doesn't have a personal photo or video... who doesn't have a camera... how many FPS of images does the average person consume AI + Image literally has the potential to replace video games. That alone…
And yet the "cursor" for images doesn't exist
❤4👾1
Ok why does tele block convex site urls ???????
I just spent HOURS debugging SSR errors
finally got it to work by using my phones hotspot
I just spent HOURS debugging SSR errors
finally got it to work by using my phones hotspot
Nothing beats paper!!! Just write out everything you have left on a piece of paper and get it all done in one day. This philosophy is why I stopped using Figma. And now ditching Anytype/Notion for project management.
🔥5
Listening to this made me realize we’re living through an era even more exciting than the early internet era
YouTube
Andrej Karpathy — “We’re summoning ghosts, not building animals”
The Andrej Karpathy episode. During this interview, Andrej explains why reinforcement learning is terrible (but everything else is much worse), why AGI will just blend into the previous ~2.5 centuries of 2% GDP growth, why self driving took so long to crack…
❤1
I am still bullish on nanobanana but when someone cooks this hard you just can't ignore it
z.ai the most exciting ai startup in China
Time to have some fun
z.ai the most exciting ai startup in China
Time to have some fun
A girl with long straight hair in 80’s anime vintage style
- Left nanobanana
- Right Bytdance dreamina-v3 (30% cheaper)
I am soooo fuckin hyped
🔥2
will to code
I am still bullish on nanobanana but when someone cooks this hard you just can't ignore it z.ai the most exciting ai startup in China Time to have some fun
I thought they had a SOTA image gen competitor but they have an 8 month old 6B model. Their main focus is coding.
So moving on to
- dreamina-v3
- hunyuan-image-3
So moving on to
- dreamina-v3
- hunyuan-image-3