Jan Bauer

14 posts

Jan Bauer banner
Jan Bauer

Jan Bauer

@japhba

PhD student at @GatsbyUCL and @ELSCbrain. Studying the mechanisms behind of intelligence 🧠🤖

London Katılım Nisan 2016
162 Takip Edilen66 Takipçiler
Jan Bauer
Jan Bauer@japhba·
@tkipf Got mine to London at the beginning of this week, perhaps it's worth reaching out again...
English
0
0
1
33
Thomas Kipf
Thomas Kipf@tkipf·
Germans living abroad: last chance to register for the upcoming election is *tomorrow* (2 Feb 2025): auswaertiges-amt.de/de/service/269…
Thomas Kipf@tkipf

If you’re German and living abroad: make sure not to miss the deadline for voter registration for the upcoming election. Even if you registered for past elections, you *will probably have to re-register* (deadline: 21 days before the election). The registration form will ask you to provide your past two registered home addresses in Germany (with exact registration/de-registration dates; thx German bureaucracy). As of September 2024, you can now submit the registration form via e-mail, see #content_2" target="_blank" rel="nofollow noopener">auswaertiges-amt.de/de/2441404-244… — it’s probably a good idea to get in touch with your local city administration ahead of time if you’re from a small town, since they might not be familiar with this new process.

English
2
7
31
10.5K
Jan Bauer
Jan Bauer@japhba·
Had a lot of fun thinking about the neural basis of cognitive flexibility with Alex, Kai, and Ali! Excited to see what other work bridging ML and Cognition will come out of the Analytical Connections summer school 🤖🧠
Alexandra Proca@a_proca

Thrilled to share our NeurIPS Spotlight paper with @japhba* @akaijsa* @SaxeLab @summerfieldlab @Ahummos*! We study how task abstractions emerge in gated linear networks and show that these abstractions support cognitive flexibility. openreview.net/pdf?id=AbTpJl7…

English
0
2
8
584
Jan Bauer
Jan Bauer@japhba·
pip ⊂ conda ⊂ docker Choose the minimum feasible isolation, but no less (Lightweight but Brittle vs Involved but Stable).
English
0
0
0
155
Jan Bauer retweetledi
Thomas Kipf
Thomas Kipf@tkipf·
Couldn't agree more: we still don't have good solutions for scalable abstraction/concept learning from "raw" data (be it language, vision, or other sensory data). Solving this would likely unlock important new capabilities.
⿻ Andrew Trask@iamtrask

For anyone interested in future LLM development One of the bigger unsolved deep learning problems: learning of hierarchical structure Example: we still use tokenizers to train SOTA LLMs. We should be able to feed in bits/chars/bytes and get SOTA Related: larger context window

English
4
15
117
29.5K
Jan Bauer retweetledi
Kording Lab 🦖
Kording Lab 🦖@KordingLab·
We often discuss if the brain does something like gradient descent. Here is me discussing the issue for @OpenNeuroMorph. I focus on exposing the weaknesses of the hypothesis as well. Should be particularly useful for newly interested people. youtube.com/watch?v=E5hATe…
YouTube video
YouTube
English
3
25
83
8.7K
Jan Bauer retweetledi
Prof. Karl Lauterbach
Prof. Karl Lauterbach@Karl_Lauterbach·
Ich bin ⁦@rezomusik⁩ Fanbase. Heute gemeinsames Video gedreht. Die Antwort auf Philip A… Kommt bald.
Prof. Karl Lauterbach tweet media
Deutsch
649
218
9.1K
1.1M
Jan Bauer retweetledi
Peyman Milanfar
Peyman Milanfar@docmilanfar·
never seen a graphic so wrong, but so useful
Peyman Milanfar tweet media
English
23
107
1.1K
123.6K
Jan Bauer retweetledi
Andrej Karpathy
Andrej Karpathy@karpathy·
@gdb 💯 reminds me of MAML meta-learning (arxiv.org/abs/1703.03400) where the objective is to find weights of a network such that any new task finetunes fast. In Software 1.0 land, equivalent is writing code such that any new desired functionality is simple and doesn't need a refactor.
English
7
10
140
33.8K
Jan Bauer
Jan Bauer@japhba·
@AlexanderDerve @GaryMarcus @MetaAI @ylecun Seems like people were expecting to handle a nail with a screwdriver. Let alone the ability to run your work through the model to include missing references seems valuable.
English
0
0
0
0
Alexander Derve
Alexander Derve@AlexanderDerve·
@GaryMarcus @MetaAI @ylecun or maybe it was removed because people like you abused the model and misrepresented it. Thanks for getting a useful and interesting public demo removed, this is why we can't have nice things.
English
29
7
202
0