zed.dev/jobs/windows-lead
zed.dev/jobs/windows-lead
With every update, we’re aiming to make the Internet less toxic, more fun, and more in your control.
• 3 minute videos!
• A separate request inbox for DMs from unknown users
• Mute accounts more quickly
With every update, we’re aiming to make the Internet less toxic, more fun, and more in your control.
• 3 minute videos!
• A separate request inbox for DMs from unknown users
• Mute accounts more quickly
A(q, K, V) = V @ softmax(K / √d @ q)
Weights: K / √d and V
nonlinearity: softmax
💡This offers fresh insights into KV cache compression research 🧵(1/3)
A(q, K, V) = V @ softmax(K / √d @ q)
Weights: K / √d and V
nonlinearity: softmax
💡This offers fresh insights into KV cache compression research 🧵(1/3)
Do not start with Machine Learning.
Do not start with Machine Learning.
In 2000, the photovoltaic power installed in the world was 385 MW and it is estimated that by 2040, this value will reach 4240 GW.
🧪🔌☀️
In 2000, the photovoltaic power installed in the world was 385 MW and it is estimated that by 2040, this value will reach 4240 GW.
🧪🔌☀️
#AcademicSky #MathSky #SciSky 🧪
#AcademicSky #MathSky #SciSky 🧪
Se você está lendo isso, você é um dos primeiros 10 milhões de usuários do Bluesky!
Se você está lendo isso, você é um dos primeiros 10 milhões de usuários do Bluesky!
Here’s a great example: eieio.games/essays/the-s...
Here’s a great example: eieio.games/essays/the-s...