Sup*

2.5K posts

Sup*

@mostlyaboutai

Casted a spell so only cool people see this profile. Did it work? I post about AI, evals, hackathon projects, vibe coding and things that make me laugh.

India Katılım Ağustos 2024

433 Takip Edilen109 Takipçiler

Sabitlenmiş Tweet

Sup*@mostlyaboutai·4 Nis

x.com/i/article/2039…

ZXX

1.2K

Sup*@mostlyaboutai·1d

@perplexity_ai Okay 👀

English

Perplexity@perplexity_ai·1d

The benchmarks show the gap. NVLS all-reduce latency drops from 586.1µs on H200 to 313.3µs on GB200. In MoE prefill at EP=4, combine falls from 730.1µs to 438.5µs. For decode, GB200 sustains much higher throughput at high token speeds.

English

11.1K

Perplexity@perplexity_ai·1d

We published new research on how we serve post-trained Qwen3 235B models on NVIDIA GB200 NVL72 Blackwell racks. GB200 is a major step up over Hopper for high-throughput inference on large MoE models, not just a training platform.

English

103

1.1K

131.6K

Sup*@mostlyaboutai·1d

All you need to do is 'imagine'

English

Sup*@mostlyaboutai·1d

This or that? Go and vote on @contra on.contra.com/mnLHDQ

English

Sup*@mostlyaboutai·2d

Just wow!

Vox@Voxyz_ai

best AI video i've seen lately. creator says they burned around $420 in tokens, model is seedance 2.0. a year ago that wouldn't even cover a 5-second demo. the speed is unreal.

English

Sup* retweetledi

🚨 AI News | TestingCatalog@testingcatalog·3d

We will likely see a deeper integration between Codex and ChatGPT already very soon. > Use the ChatGPT app on your phone to keep working with Codex whenever your computer is awake. Additionally, this image from OpenAI sparked loads of speculations, including the one where OpenAI would be teasing their own mobile phone. Even though it is quite unrealistic, this would be a huge steal of attention from the Google I/O event.

\\@zyrnob

how do i call x.com/sama/status/20…

English

936

134K

Sup*@mostlyaboutai·3d

@sundeep That's the thing with resources, though abundant, they are not equally distributed. AI and Tech too, are resources just like any other resource.

English

sunny madra@sundeep·3d

Accurate.

Elad Gil@eladgil

People at major AI labs (using internal models) 3-4 months ahead of startup silicon valley engineers SV founders/eng 3-6 months ahead of NY NY founders/eng 6-12 months ahead of rest of world Most people have no idea how fast AI shifting as 1-2 years behind SOTA "The future is here, just not equally distributed" - Robert Heinlein

English

6.1K

Sup*@mostlyaboutai·3d

@marcelkargul @pizzaboy Yeah they are a pretty big agency. Funny how the 'again' indicates that this ain't the first time they are stealing his work 👀

English

Marcel@marcelkargul·3d

@pizzaboy huge agency on dribbble especially

Victa Wille@VictaWille

Just look at this cringe. These guys stole my work again — and it's not the first time. They didn’t even bother to change anything. Do you have a problem, guys? @qclay_design @bogdan_qclay @awsmd

English

402

Dan@pizzaboy·3d

Time to start making a list x.com/i/lists/205337…

English

182

8.8K

Sup*@mostlyaboutai·3d

.....

ZXX

Sup*@mostlyaboutai·3d

@pizzaboy If I ever do, I will self report haha. Btw this is a great initiative.

English

Dan@pizzaboy·3d

@mostlyaboutai I'm adding you

English

Sup*@mostlyaboutai·3d

@pizzaboy Obv. I was being sarcastic here Dan 👀 If you have a good design team, most designs people post here on X are probably not even worth stealing anyway.

English

Dan@pizzaboy·3d

@mostlyaboutai Don’t steal then

English

136

Sup*@mostlyaboutai·3d

Model Centric Prompt Layman's Prompt Prompt written using GPT

Hugeicons@huge_icons

too relatable?😅

English

Sup*@mostlyaboutai·3d

They are basically apple but with wheels Nvm, I will buy one very soon 👀

Lamborghini@Lamborghini

We are proud to introduce Lamborghini Fenomeno Roadster: the most powerful open-top ever created by Lamborghini. Limited to 15 units, powered by an iconic 1080 CV naturally aspirated V12 hybrid HPEV. Engineered with an aerospace-inspired carbon fiber monofuselage, advanced active aerodynamics, and capable of 0–100 km/h in just 2.4 seconds, Fenomeno Roadster delivers pure performance with uncompromising driving emotion. A V12 symphony. #Lamborghini

English

Sup*@mostlyaboutai·3d

@caleb_friesen Dude you legit have no idea of the kind of work we have done for Indian Startups 👀

English

Caleb@caleb_friesen·4d

Name an Indian tech co with cooler swag than Sarvam. I'll wait.

Jonas@jonaswillett1

what tech company / startup has the best merch? 👀

English

1.7K

92.8K

Sup*@mostlyaboutai·3d

I have been influenced by the concept of rating models based on Taste so I am conducting a little experiment of my own. AI can get you almost anything you want. Ask for slop, and slop is exactly what you’ll receive. The real fun begins when the prompts land perfectly and the result comes out absolutely stunning.

English

Sup*@mostlyaboutai·3d

@brotzky We are all sailing the same ship, and it even has a cute name... "Titanic."

English

632

Brotzky@brotzky·3d

Accidentally opened Cursor and holy shit it was weird to look at code and a file system. How times change.

English

210

23.1K

Sup*@mostlyaboutai·3d

@sama Less Hallucination Increased Confidence - 5.5 is easy to break and take in false info. Better code structuring Open Source (Idk but I am gonna keep this here) Ability to use the research it does.

English

120