Ingo Marquardt
banner
ingo-m.bsky.social
Ingo Marquardt
@ingo-m.bsky.social
Applied Scientist | Ex-Neuroscientist | Currently @restack.io | ingomarquardt.io
Unless someone comes up with a new, effective way to prop up the company through massive political influence
April 17, 2025 at 10:10 PM
Is there a way to programmatically identify the top 100 or so ML related posts on twitter and repost them here, so we don't have to open twitter anymore? Is there still any API access?
March 2, 2025 at 1:42 AM
Also, another unstable 3 party coalition will decrease trust in democracy, it is the road to a fascist government in the next election.
February 23, 2025 at 7:22 PM
Looks like either FDP or BSW gets in, in that case a 3 party coalition will be necessary. That means instability. It's terrifying.
February 23, 2025 at 7:21 PM
In liberal democracies, exploitation and violence is a bug. In authoritarianism, it's a feature.
February 18, 2025 at 3:05 PM
Exploitation & aggression (at home and abroad) does exist in liberal democracies. But with liberal democracy, there's at least a chance to overcome it. In authoritarianism, exploitation is guaranteed.
February 18, 2025 at 3:05 PM
I agree it sounds bad. But we have to realize that to defend its interests (liberal democracy), Europe has to wield power. Being on the right side of history won't help us when facing authoritarian adversaries.
February 18, 2025 at 12:20 PM
Yes, fascism will be defeated again, as it has been before, and that should give us hope, but the grim question is what damage will be done until then.
February 17, 2025 at 1:09 AM
Would be great if someone posted daily summaries of hot ML topics from twitter here. I'm off twitter and probably missing some relevant stuff, but I don't want to open that app again.
February 17, 2025 at 12:52 AM
I'd love to see 10 learning sins (that were actually a good idea) 😅
February 6, 2025 at 1:28 PM
Many people bought Teslas in the past for good reasons and do not support Elon
February 4, 2025 at 12:01 PM
Ok that's an interesting take I wasn't aware of. As someone with a background in in neuroscience I have a bias in favour of latent continuous. But I'll admit it's more of a feeling than a rigours argument.
January 26, 2025 at 1:39 AM
This phone call didn't raise any concerns with his supporters so I don't think anything ever will youtu.be/AW_Bdf_jGaA?...
Full Phone Call: Trump Pressures Georgia Secretary of State To Recount Election Votes | NBC News
YouTube video by NBC News
youtu.be
January 25, 2025 at 2:43 AM
Isn't deepseek R1's reasoning in token space?
January 25, 2025 at 2:33 AM
Is it known whether OpenAI's o3 "private chain of thought" is entirely in token space, or whether they use a latent space technique as well? 🤔

[1] arxiv.org/abs/2412.06769
[2] arxiv.org/abs/2412.17747
Training Large Language Models to Reason in a Continuous Latent Space
Large language models (LLMs) are restricted to reason in the "language space", where they typically express the reasoning process with a chain-of-thought (CoT) to solve a complex reasoning problem. Ho...
arxiv.org
January 13, 2025 at 7:31 PM
Ubuntu LTS (currently version 24) with gnome (i.e. no modifications). Why? Because I need to get work done and it's solid. Just works. Have been using it for many years. Previously I was a mac user.
January 10, 2025 at 12:09 AM
Call it "brute force" or not, but "thousands" of USD per task for a task that's easy for humans is not ideal. Perhaps this is progress (we don't know due to lack of transparency). But not the solution many were hoping for.
December 21, 2024 at 9:59 AM