Hao Liu

Deep learning scientist at Jusfoun. ex-SQE at Redhat. PhD from CAS in theoretical & computational chemistry.

[Hao Liu @liuhaonk]

@topazlabs Looks cool, Can’t wait to try it out.

[Hao Liu @liuhaonk]

@pickover Resell the green button to others as an Instant 10M botton.

[Eli Dourado @elidourado]

New blog post:

Is the Great Stagnation ending?
What technologies am I watching in the decade ahead?
Are we going to get life extension treatments soon?

5700+ words on all of this. Please read and share!

https://t.co/5zIE48XpIC

[Mark Lewis, MD, FASCO @marklewismd]

Yikes, would you take a drug* with this side effect profile?!

(*oh, it's the placebo arm of the Pfizer vaccine trial)

[Thomas Cabaret (thomascabaret84.bsky.social) @ThomasCabaret84]

En blanc: comment mettre 30 briques sur un rebord pour maximiser le déport (avec 1 brique par étage comme contrainte)
En gris: la même chose sans la contrainte.

[Dominique Beaini @dom_beaini]

@xkcdComic If your model doesn't work, simply optimize the parameters!

[Elisabeth Bik @MicrobiomDigest]

I am ringing the alarm.
We have now found >400 papers that all share a very similar title layout, graph layout, and (most importantly) the same Western blot layout.
This is a massive #PaperMill of (what we assume) fabricated data.

[Lionel Page @page_eco]

This is a black and white photograph. Only the lines have colour.

What you “see” is what your 🧠 predicts the reality to be, given the imperfect information it gets.

[algoritmic @algoritmic]

Understanding Apache Kafka with Factorio https://t.co/NuJRnyUlym

[Marcos López de Prado @lopezdeprado]

A common misconception is that the risk of overfitting increases with the number of parameters in the model. In reality, a single parameter suffices to fit most datasets: https://t.co/4eOGBIyZl9

Implementation available at: https://t.co/xKikc2m0Yf

[Daniel Sinclair 📍 SF @DanielSinclair]

Here's a video of a Chimpanzee using Instagram. It's eye-opening. https://t.co/FEfIgQSGa6

[monkeyUser @ismonkeyuser]

Curiosity (https://t.co/vVOAtmZgJO)

[algoritmic @algoritmic]

HTML/CSS drawing in the style of an 18th-century oil painting. Hand-coded entirely in HTML & CSS https://t.co/eroAFOYxP6 by @cyanharlow

[Greg Brockman @gdb]

An OpenAI employee printed out this AI-written sample and posted it by the recycling bin: https://t.co/PT8CMSU2AR

[Ricard Solé @ricard_sole]

Two types of brains have emerged in nature: solid (ours) and liquid (ant colonies, immune system). What can be computed by liquid brains? How can they learn? Are there universal rules? Here's our new paper on the statistical physics of liquid brains https://t.co/ZdyUnCaoiW

[Andrej Karpathy @karpathy]

Differentiable Monte Carlo Ray Tracing through Edge Sampling https://t.co/NtVmZUxkz9 "differentiable ray tracer [...] able to compute derivatives of scalar functions over a rendered image w.r.t. parameters such as camera pose, scene geometry, materials, and lighting" fun!

[Andrew Hart @AndrewHart]

Here’s a peek at what I’ve been working towards for the past few months. Introducing Retail AR, from Dent Reality. https://t.co/CpS0w3ZXYX

[Misha Ahrens @MishaAhrens]

We made public:

1.5 million neurons, cell-level activity during multiple behaviors
15 TB raw, 56 GB compressed
18 fish whole-brain data, ~100k neurons/fish

Data and code: https://t.co/KO2IEabdV6
Paper: https://t.co/2H6l3zWrRL

[Greg Brockman @gdb]

Performance on Montezuma's Revenge, on which the 2015 DQN algorithm famously achieved a score of 0. We've now developed an algorithm which exceeds human performance: https://t.co/lGw3rZUbv3