13.03.2026 03:14
👍 0
🔁 1
💬 0
📌 0
Im pretty sure there is a YouTube video/article/something about this very specific example and how it got lost in translation. But I don’t remember
I’ve always liked the onion’s eyes in this one
www.percepta.ai/blog/can-llm...
As a research lark at Percepta, Christos embedded a computer into an LLM, showed that it could solve the hardest Sudokus, and then as a side bonus built an exponentially faster attention
Top Panel: Shows Gary (Jason Segel), Mary (Amy Adams), and the Muppet Walter standing together with concerned and slightly confused expressions. They are labeled "Boomers," "Gen Z," and "Gen Alpha," respectively. Bottom Panel: Shows a dramatic, low-angle shot of Kermit the Frog's silhouette as he walks forward, carrying a small bag. It is captioned: "Millennials coming to fix the computer."
Sorry but this is far too accurate
Coding agents aside, a 30B valuation for a VScode fork should have raised more flags.
vai brazil!!!
Because you shouldn’t use them
We are not at war. Israel is.
You know the rule. Confirming evidences of you riding or the article is wrong
LOL so hard: Spot a subtle architectural flaw in a 1000 line PR
What if I wanted sorted by last name. Skipping the years
Ultraparameters you mean
*Brasil
one aspect where you probably beat Mr Beast
Rad
Only after it turns 16!? 😁
You believe this is why Amazon fired them
How strictly do you/the book adhere to TDD?
How are you doing reactivity? thought about using Marimo instead?
Or just observe that peanut allergy is not really a thing in any other country.
I guess my main issue was with the akas: “reasoning" aka inference-scaling aka Reinforcement Learning from Verifiable Rewards (RLVR)”
They are not as know as
No they are independent. A model can be trained with RLVR, but users can use thinking mode or not.
Their goal is the same tho: correct answers. But how and when they achieve that differs
And when they take place. RLVR is during training, reasoning at inference time
Reasoning can be used to any and every topic. RLVR only to very specific topics that can be verified by external means. I can’t think anything else besides math, coding and puzzles like Andrej mentioned
Se tiver mais que um?? Tira a média ponderada?
Don’t quite agree with the first topic. These are 3 related things, not the same thing. Reasoning, RLVR and inference scaling
Boa noite Nobre. Até ano que vem
another robot highlight for 2025: man wearing humanoid mocap suit kicks himself in the balls
