stoXe

I make things work & I break them. Depends what the day calls for. Tweets and opinions are my own.

[stoXe @DevinStokes]

@dcuthbert You..... lol

[stoXe @DevinStokes]

@lukedepulford orly

[stoXe @DevinStokes]

Infinite Likes On Twitter:

1. Create tweet.
2. Inspect element and get CSS Selector for the like button
3. `setInterval(() => {
document.querySelector("SELECTOR_HERE")?.click();
}, 100);`
4...
6. Watch the likes increase.
7. No profit because its largely client side :)

[stoXe @DevinStokes]

I know some of these words!

https://t.co/qsyIeyr8N1

[wordgrammer @wordgrammer]

Q: How did they replicate o1?

A: Reinforcement learning. Take complicated questions that can be easily verified (either math or code). Update the model if correct.

[stoXe @DevinStokes]

"I'd love to learn how language models work!"

oh...

[wordgrammer @wordgrammer]

Q: How did DeepSeek train so much more efficiently?

A: They used the formulas below to “predict” which tokens the model would activate. Then, they only trained these tokens. They need 95% fewer GPUs than Meta because for each token, they only trained 5% of their parameters.

[stoXe @DevinStokes]

To me that suggests an inherent weight (unintended pun) given to its "intrinsic" knowledge, as opposed to what is presented during the reasoning.

[stoXe @DevinStokes]

Cool new model - DeepSeek R-1 utilizes a Chain of Thought style similar to o1 from OpenAI.

However I am not convinced it utilizes the reasoning in a meaningful way when its inherent "knowledge" conflicts with its reasoning steps.

[stoXe @DevinStokes]

1. During its counting process, it repeatedly finds 3 "r"s (at positions 3, 8, and 9)

2. However, its intrinsic knowledge that "strawberry" has "two Rs" keeps overriding this direct evidence

3. This suggests weight given to the LLM's intrinsic knowledge

[stoXe @DevinStokes]

@_mattata @vxunderground cat: /dev/input/event: No such file or directory

[stoXe @DevinStokes]

@_mattata @vxunderground Wait wut

[stoXe @DevinStokes]

@notdan @Microsoft mY fAvOrItE

[stoXe @DevinStokes]

Drops remarkable text-to-video model.

Has 140+ followers.

This is the tech I love.

[Yoav HaCohen @yoavhacohen]

Excited to announce LTX-Video!

Our new text-to-video model generates stunning, high-quality videos faster than real-time—5 seconds of 24fps video at 768x512 in just 4 seconds on an Nvidia H100! ⚡

We’re open-sourcing the code & weights. Check out the results 🎥👇

[stoXe @DevinStokes]

Relativty - An open-source VR headset
https://t.co/W2xZgaiPqx

[stoXe @DevinStokes]

@MasterTimBlais curl

[stoXe @DevinStokes]

Is this intentional attemp at flying under the "radar" of tech news?

They send out a warning of RCE, but don't even assign a CVE?

https://t.co/nYvlgQzUgP

[stoXe @DevinStokes]

@gabsmashh Felt cute. Might decrypt later.

[stoXe @DevinStokes]

@TomStSyber @UK_Daniel_Card I think that is referring to the standard models. They are touch capacitive, not finger print.