spectraldani's profile picture. PhD student at UCL. Working on Gaussian Processes. I love everything mathy, videogames, and language. 🇧🇷
He/Ele/Él

@spectraldani@sigmoid.social

Daniel Augusto

@spectraldani

PhD student at UCL. Working on Gaussian Processes. I love everything mathy, videogames, and language. 🇧🇷 He/Ele/Él @[email protected]

Pinned

Happy to have joined this great team :) In which circunstances can we use abundant proxy preferences to quickly learn true preferences? I'm glad to announce our paper explores and proposes a model for one of these cases. Follow @_zhuyuchen for more details. #ICML2025

spectraldani's tweet image. Happy to have joined this great team :)
In which circunstances can we use abundant proxy preferences to quickly learn true preferences? I'm glad to announce our paper explores and proposes a model for one of these cases.  Follow @_zhuyuchen for more details. #ICML2025

New work! 💪🏻💥🤯 When Can Proxies Improve the Sample Complexity of Preference Learning? Our paper is accepted at @icmlconf 2025. Fantastic joint work with @spectraldani @zhengyan_shi @Mengyue_Yang_ @PMinervini @alexdamour, Matt Kusner. 1/n



Daniel Augusto reposted

A short blogpost on 𰻞/𰻝 (from the famous biang-biang noodles) and its relatives. I briefly described their origin, spread, the variability of their components, and the emergence of their readings and meanings. Lots of pictures too, enjoy. blog333610347.wordpress.com/2025/11/08/bia…


Daniel Augusto reposted

TL;DR: I made a Transformer that conditions its generation on latent variables. To do so an encoder Transformer only needs a source of randomness during generation, but then it needs an encoder for training, as a [conditional] VAE. 1/5


Daniel Augusto reposted

A PhD can be incredibly rewarding and fun, under the right circumstances. If you have tremendous *fire* for research and self motivation, it can be the right path for you. If you're doing it because it's the least friction next step after ugrad or MS, reconsider.

I still get a lot of questions about doing a PhD in ML so I'm resurfacing a three threads I wrote about this including some fun personal stories 🧵



Daniel Augusto reposted

lmao Discord just suffered a breach in which hostile actors stole people's govt IDs if they handed them over to Discord Support and the govt and big tech still wants you to hand over your ID for age verification just so you can browse the web. it's a fucking recipe for disaster

This post is unavailable.

Daniel Augusto reposted

Gödel proved in 1932 that intuitionistic logic has no interpretation as a finite-valued logic. The proof was only two pages long.


Daniel Augusto reposted

とりゃ〜

nicoberu12's tweet image. とりゃ〜

Daniel Augusto reposted

What I’ve been up to lately btw

sora_nothere's tweet image. What I’ve been up to lately btw

Daniel Augusto reposted

i've finally got a new blogpost out!! this one talks about modern CSS, it's new features, and practical real world uses as usual, the visuals are css-only and have lots of cool interactivity and easter eggs :3 have fun! lyra.horse/blog/2025/08/y…


Daniel Augusto reposted

The last digit of a year, in the Common Era/ Anno Domini corresponds to its Heavenly Stem/Tiangan in the Chinese calendar, as shown in the table below. For example,1848, 938,1988, 2008 all end in 8, and so they are years whose heavenly stem is Wu 戊

edwardW2's tweet image. The last digit of a year, in the Common Era/ Anno Domini corresponds to its Heavenly Stem/Tiangan in the Chinese calendar, as shown in the table below. 

 For example,1848, 938,1988, 2008 all end in 8, and so they are years whose heavenly stem is Wu 戊

Daniel Augusto reposted

The surge in AI-written speeches in Britain's House of Commons visualised:

Valen10Francois's tweet image. The surge in AI-written speeches in Britain's House of Commons visualised:

Tom Tugendhat calls out the rise of MPs using ChatGPT to write speeches 👏🏻👏🏻 “This place has become absurd”



Daniel Augusto reposted

Lecture 1 from this semester's "Introduction to Formal Real Analysis" is posted: youtube.com/watch?v=TN51-1… We first go over why "Real Analysis" is even needed (why is "Calculus" insufficiently rigorous), and then the first steps in using Lean. We're following "Real Analysis, The…

AlexKontorovich's tweet card. Lecture 1, Introduction to Formal Real Analysis, Rutgers University,...

youtube.com

YouTube

Lecture 1, Introduction to Formal Real Analysis, Rutgers University,...


Daniel Augusto reposted

A delegada liderou a trágica operação Ouvidos Moucos. Acusaram o reitor da UFSC de desviar R$ 80 milhões - que, a rigor, era o valor total do programa de ensino à distância Luiz Carlos Cancellier era inocente. A única "prova" contra ele: depósitos na conta do filho no valor de…

Com 50 anos de idade.

pfnery's tweet image. Com 50 anos de idade.


Daniel Augusto reposted

Looking at the thread. The common frame to look at the more general phenomenon involves an eigenproblem of the form Oƒ = λƒ, where the operator O encodes either: a symmetry (translations, rotations, general group transformations), or a a statistic (e.g. covariance, correlation),

i just realized PCA is Fourier analysis without the sines and cosines, but with your data’s own modes.



Daniel Augusto reposted

Today I scrolled through Daikanwa, and there was one thing that catched my eye. It was the 西夏語 "Tangut language" word out of nowhere. So I think it's time for a short thread. /1

gyankotsu's tweet image. Today I scrolled through Daikanwa, and there was one thing that catched my eye. It was the 西夏語 "Tangut language" word out of nowhere. So I think it's time for a short thread. /1

Daniel Augusto reposted

WTF?! NeurIPS sneakily edited their website AFTER the deadline to say papers whose authors only have virtual passes will be desk rejected, and in person registration is needed Why do all the conferences want to screw over disadvantaged researchers? They're already rich!

NeelNanda5's tweet image. WTF?! NeurIPS sneakily edited their website AFTER the deadline to say papers whose authors only have virtual passes will be desk rejected, and in person registration is needed

Why do all the conferences want to screw over disadvantaged researchers? They're already rich!
NeelNanda5's tweet image. WTF?! NeurIPS sneakily edited their website AFTER the deadline to say papers whose authors only have virtual passes will be desk rejected, and in person registration is needed

Why do all the conferences want to screw over disadvantaged researchers? They're already rich!

Daniel Augusto reposted
XianyangCB's tweet image. Counterpoint:  ctext.org/dictionary.pl?…

The ancients aspired to the Way, held fast to virtue, and wandered in the arts—ritual, music, archery, charioteering, writing, and arithmetic. Arithmetic is particularly the least important matter. – Zhu Xi, Yulei 14.B8 古人志道,據德,而游於藝:禮樂射御書數,數尤為最末事。



Daniel Augusto reposted

今日の目標💻

esusansan's tweet image. 今日の目標💻

Daniel Augusto reposted

there's an interesting book called _the art of not being governed_ that touches the phenomena of tribes converting en masse to christianity. To summarise its argument, these tribes are not primitive people that were somehow left behind by civilisation, but people...


Daniel Augusto reposted

The #AISTATS2026 Call for Papers is out! Abstract Deadline: Sep 25, 2025 (AoE) Full paper Deadline: Oct 2, 2025 (AoE) Supplementary Deadline: Oct 9, 2025 (AoE) It is a privilege to serve as a Workflow Chair alongside @_Mr_YJ and Program Chairs Aaditya Ramdas and @arnosolin


Daniel Augusto reposted

Super happy to share the news on behalf of the team for the Sally T. WongAvery Digital Collection of Chinese Materials at Cambridge University Library: our Hopkins Oracle Bones (2D set CUL1–610) and 50 3D images are now freely accessible to the world. cudl.lib.cam.ac.uk/collections/wo…

YanHeCUL's tweet image. Super happy to share the news on behalf of the team for the Sally T. WongAvery Digital Collection of Chinese Materials at Cambridge University Library: our Hopkins Oracle Bones (2D set CUL1–610) and 50 3D images are now freely accessible to the world. cudl.lib.cam.ac.uk/collections/wo…
YanHeCUL's tweet image. Super happy to share the news on behalf of the team for the Sally T. WongAvery Digital Collection of Chinese Materials at Cambridge University Library: our Hopkins Oracle Bones (2D set CUL1–610) and 50 3D images are now freely accessible to the world. cudl.lib.cam.ac.uk/collections/wo…
YanHeCUL's tweet image. Super happy to share the news on behalf of the team for the Sally T. WongAvery Digital Collection of Chinese Materials at Cambridge University Library: our Hopkins Oracle Bones (2D set CUL1–610) and 50 3D images are now freely accessible to the world. cudl.lib.cam.ac.uk/collections/wo…
YanHeCUL's tweet image. Super happy to share the news on behalf of the team for the Sally T. WongAvery Digital Collection of Chinese Materials at Cambridge University Library: our Hopkins Oracle Bones (2D set CUL1–610) and 50 3D images are now freely accessible to the world. cudl.lib.cam.ac.uk/collections/wo…

Loading...

Something went wrong.


Something went wrong.