πŸ§ͺπŸ–ŒπŸͺ„'s Avatar

πŸ§ͺπŸ–ŒπŸͺ„

@scienceartmagic.bsky.social

Open Source AI/ML Research https://github.com/ScienceArtMagic https://hf.co/ScienceArtMagic https://www.youtube.com/@ScienceArtMagic

368 Followers  |  3,827 Following  |  175 Posts  |  Joined: 17.11.2023  |  1.7556

Latest posts by scienceartmagic.bsky.social on Bluesky


Post image 04.02.2026 03:42 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Jeffrey Epstein email sharing advice from Noam Chomsky to Epstein.

Jeffrey Epstein email sharing advice from Noam Chomsky to Epstein.

AI/ML folks:

Can we finally stop pretending that pseudo-intellectual, pro-Putin mouthpiece (and apparently pedo advisor even if not one himself - which is highly unlikely given his ties) Noam Chomsky has *anything* to offer the ML world?

02.02.2026 18:16 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Forcing PhD students to publish before thesis submission created a predatory journal gold rush.

Too many couldn't get into real journals fast enough.

So they paid to publish garbage.

Quality PhD work produces quality papers.

Not the other way around.

04.01.2026 01:03 β€” πŸ‘ 3    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0
Twitter user @joni_askola: "The most overestimated IQ of the 21st century" (referring to photo of Elon Musk)

Twitter user @joni_askola: "The most overestimated IQ of the 21st century" (referring to photo of Elon Musk)

28.12.2025 09:28 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

I dunno. I find it completely tenable to continue to dismiss any claims of consciousness for large language models.

08.12.2025 20:28 β€” πŸ‘ 47    πŸ” 5    πŸ’¬ 2    πŸ“Œ 7

NotebookLM's lack of Google Keep integration is the most Google thing ever

26.10.2025 03:10 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Preview
John Stockton admits he is not watching the NBA anymore because it's way too soft: "Fans want you to go out there and do what they can't do, not go out there, shake hands and hug" Hall of Famer John Stockton says load management, softer competition and the lack of true rivalries have drained his interest in watching today’s NBA.

Biggest flopper in #NBA history says modern NBA is too soft:

sports.yahoo.com/article/john...

#NBAsky

15.08.2025 07:04 β€” πŸ‘ 4    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Slightly diminish a band:

Power Walk The Jewels

13.08.2025 03:24 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

When Claude responds with "You're absolutely right," humans bemoan LLM sycophancy.

Humans when queried by anyone who might help promote their paper or fund their product/connect them to funding:
"That's a great question..."

06.08.2025 17:28 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

JFC this is gonna be ugly

03.08.2025 21:40 β€” πŸ‘ 178    πŸ” 43    πŸ’¬ 7    πŸ“Œ 13

Humans when reading "delve," "multifaceted," em dash, etc.:
"AI slop."

Humans any time ML academia, investors, or just some random techbro asks something:
"That's a great question..."

14.07.2025 22:53 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 1    πŸ“Œ 1

Any politician who runs on banning the phrase "that's a great question" will have my vote.

14.07.2025 22:41 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 1
MCP-B: A Protocol for AI Browser Automation Posted by bustodisgusto with a score of 98

MCP-B: A Protocol for AI Browser Automation

view on hacker news

10.07.2025 02:00 β€” πŸ‘ 1    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0
Person sitting with a banner that reads YOU ARE ALL COWARDS

Person sitting with a banner that reads YOU ARE ALL COWARDS

This guy is just posted up outside of Congress and I’m honestly not sure what his political views are but I agree

13.06.2025 12:45 β€” πŸ‘ 21193    πŸ” 2886    πŸ’¬ 217    πŸ“Œ 158

Saw video on Twitterchan of a loud cheers for the drag queens, and a mix of cheers and (mostly) boos for pedo king even in a venue he tried to pack with MAGAts.

12.06.2025 05:55 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

Chengpeng Li, Zhengyang Tang, Ziniu Li, Mingfeng Xue, Keqin Bao, Tian Ding, Ruoyu Sun, Benyou Wang, Xiang Wang, Junyang Lin, Dayiheng Liu
CoRT: Code-integrated Reasoning within Thinking
https://arxiv.org/abs/2506.09820

12.06.2025 05:51 β€” πŸ‘ 1    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0
Preview
Are Optimal Algorithms Still Optimal? Rethinking Sorting in LLM-Based Pairwise Ranking with Batching and Caching We introduce a novel framework for analyzing sorting algorithms in pairwise ranking prompting (PRP), re-centering the cost model around LLM inferences rather than traditional pairwise comparisons. Whi...

Are Optimal Algorithms Still Optimal? Rethinking Sorting in LLM-Based Pairwise Ranking with Batching and Caching

Reveals that traditional sorting algorithm efficiency rankings break down in LLM contexts.

πŸ“ arxiv.org/abs/2505.24643

02.06.2025 02:46 β€” πŸ‘ 2    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0
Preview
Pangu DeepDiver: Adaptive Search Intensity Scaling via Open-Web Reinforcement Learning Information seeking demands iterative evidence gathering and reflective reasoning, yet large language models (LLMs) still struggle with it in open-web question answering. Existing methods rely on stat...

Pangu DeepDiver: Adaptive Search Intensity Scaling via Open-Web Reinforcement Learning

Huawei introduces an RL framework that enables LLMs to dynamically scale search frequency and depth based on information demands in real-world web environments.

πŸ“ arxiv.org/abs/2505.24332

02.06.2025 02:46 β€” πŸ‘ 1    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0
Preview
GitHub MCP Exploited: Accessing private repositories via MCP We showcase a critical vulnerability with the official GitHub MCP server, allowing attackers to access private repository data. The vulnerability is among the first discovered by Invariant's security ...

invariantlabs.ai/blog/mcp-git...

27.05.2025 14:58 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Preview
Nonlinear Meta-Learning Can Guarantee Faster Rates Many recent theoretical works on \emph{meta-learning} aim to achieve guarantees in leveraging similar representational structures from related tasks towards simplifying a target task. The main aim of ...

🚨 New paper accepted at SIMODS! 🚨
β€œNonlinear Meta-learning Can Guarantee Faster Rates”

arxiv.org/abs/2307.10870

When does meta learning work? Spoiler: generalise to new tasks by overfitting on your training tasks!

Here is why:
πŸ§΅πŸ‘‡

26.05.2025 16:50 β€” πŸ‘ 9    πŸ” 7    πŸ’¬ 2    πŸ“Œ 1
Preview
Fender just launched its own free DAW software for recording music Iconic brand Fender just launched Fender Studio. This free music-recording software is primarily intended for newbies.

I did not have "Fender releases a free DAW" on my bingo card, let alone "for Windows, Mac, iOS, Android, and LINUX."

www.engadget.com/audio/fender...

20.05.2025 23:21 β€” πŸ‘ 5    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0

David Anugraha, Zilu Tang, Lester James V. Miranda, Hanyang Zhao, Mohammad Rifqi Farhansyah, Garry Kuwanto, Derry Wijaya, Genta Indra Winata
R3: Robust Rubric-Agnostic Reward Models
https://arxiv.org/abs/2505.13388

20.05.2025 05:58 β€” πŸ‘ 1    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0

Renjie Pi, Felix Bai, Qibin Chen, Simon Wang, Jiulong Shan, Kieran Liu, Meng Cao
MR. Judge: Multimodal Reasoner as a Judge
https://arxiv.org/abs/2505.13403

20.05.2025 05:58 β€” πŸ‘ 1    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0

Nithin Rao Koluguri, Monica Sekoyan, George Zelenfroynd, Sasha Meister, Shuoyang Ding, Sofia Kostandian, He Huang, Nikolay Karpov, Jagadeesh Balam, Vitaly Lavrukhin, ...
Granary: Speech Recognition and Translation Dataset in 25 European Languages
https://arxiv.org/abs/2505.13404

20.05.2025 05:57 β€” πŸ‘ 1    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0
Marine mystery of octopus riding shark
YouTube video by University of Auckland - Website video repository Marine mystery of octopus riding shark

Just an octopus hitching a ride on the world's fastest shark...

We need to eradicate cephalopods. 86 8 before it's too late!

youtu.be/3cYlBarkmDY

17.05.2025 18:39 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 1

Taking me down memory lane. Mine was a very kludgy impl of assemble (remember grunt, pre-gulp?) with custom markdown-it plugins that also fed to pandoc for my resume in docs format. Back when I could I could look down on Bootstrap users with my shiny new Foundation theme.

17.05.2025 00:28 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Preview
GitHub - graze-social/aip: ATmosphere Authentication, Identity, and Permission Proxy ATmosphere Authentication, Identity, and Permission Proxy - graze-social/aip

Authentication is hard. We've been juggling OAuth and pockets full of app passwords, and we're know we're not the only ones. That's why today, we're open sourcing an all in one, best-in-class service that anyone can use to deal with ATProto authentication - for free.

01.05.2025 20:02 β€” πŸ‘ 238    πŸ” 49    πŸ’¬ 12    πŸ“Œ 10
Preview
xAI Dev Leaks API Key for Private SpaceX, Tesla LLMs A employee at Elon Musk's artificial intelligence company xAIΒ leaked a private key on GitHub that for the past two months could have allowed anyone to query private xAI large language models (LLMs) wh...

krebsonsecurity.com/2025/05/xai-...

02.05.2025 07:03 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Seriously @bsky.app, you've got to figure out your deep links. Sharing from another app to Bluesky is completely broken in Android. Goes to the home (feed) screen every time.

02.05.2025 02:26 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

@scienceartmagic is following 19 prominent accounts