Sign in

tiagogiraldo.bsky.social

@tiagogiraldo.bsky.social
80 followers 488 following 36 posts
PostsRepliesMedia
Reposted by @tiagogiraldo.bsky.social
Compute! Paris @computeparis.bsky.social · 28/09/2026
At Compute! Paris, Francesco Bruzzesi introduces polars-stats, a Polars expression plugin doing exactly that, Rust (statrs) underneath, column-valued parameters as a first-class feature, sampling reproducible across platforms.
github.com
GitHub - FBruzzesi/polars-stats: Polars expression plugin that exposes `scipy.stats`-style probability distributions
Polars expression plugin that exposes `scipy.stats`-style probability distributions - FBruzzesi/polars-stats
101
Reposted by @tiagogiraldo.bsky.social
hardmaru @hardmaru.bsky.social · 28/09/2026
Our Neuroevolution textbook is finally in print! Free online edition: neuroevolutionbook.com Pre-order: a.co/d/06DpkH9h I am incredibly grateful to my co-authors Sebastian Risi, Yujin Tang, and Risto Miikkulainen for making this happen. Thread:
23913
Reposted by @tiagogiraldo.bsky.social
NBER @nber.org · 28/09/2026
An LLM workflow that reproduces, improves, and extends economics research has been applied to over 4,000 published replication packages, from Matthew Schwartz, Isaiah Andrews, and Jesse M. Shapiro www.nber.org/papers/w35782
0197
Reposted by @tiagogiraldo.bsky.social
Gradient Flow @gradientflow.com · 28/09/2026
🎗️Six Reasons I Think Open AI Models Will Win 🏆 gradientflow.com/six-reasons-...
gradientflow.com
Six Reasons I Think Open AI Models Will Win - Gradient Flow
In my conversations with developers and AI teams, I’m struck by how many are exploring moving more of their inference workloads to open models. I’ve touched on some of their reasons before , but here ...
011
Reposted by @tiagogiraldo.bsky.social
Science X / Phys.org @sciencex.bsky.social · 07/09/2026
Ambiguous digit images made competing AI brain models disagree, and purely discriminative models could not explain human judgments. Generative computations appeared to matter for visual recognition. doi.org/hchwb2
medicalxpress.com
Why scientists are looking to provoke disagreement among AI brain models
When people wanted to create better artificial intelligence, they looked to the brain for inspiration. Now the tables are turning, and researchers are using AI models to better understand the brain.
031
Reposted by @tiagogiraldo.bsky.social
Sung Kim @sungkim.bsky.social · 05/09/2026
An interesting paper. AI-Native Firms "More broadly, our findings suggest that AI may not simply make existing organizations more efficient—it may change what organizations look like and do" www.hbs.edu/ris/Publicat...
1223
Reposted by @tiagogiraldo.bsky.social
Anthony Panozzo @panozzaj.com · 05/04/2025
Interesting post: A Field Guide to Rapidly Improving AI Products by @hamel.bsky.social
hamel.dev
A Field Guide to Rapidly Improving AI Products – Hamel’s Blog
Evaluation methods, data-driven improvement, and experimentation techniques from 30+ production implementations.
032
Reposted by @tiagogiraldo.bsky.social
Gradient Flow @gradientflow.com · 06/04/2025
Accelerate research with #AI tools that autonomously gather, analyze, and cite sources → gradientflow.substack.com/p/autonomous...
gradientflow.substack.com
Autonomous AI Agents Are Changing Knowledge Work—Fast
Subscribe • Previous Issues
011
Reposted by @tiagogiraldo.bsky.social
Jacy Reese Anthis @jacyanthis.bsky.social · 04/04/2025
Should we use LLMs 🤖 to simulate human research subjects 🧑? In our new preprint, we argue sims can augment human studies to scale up social science as AI technology accelerates. We identify 5 tractable challenges and argue this is a promising and underused research method 🧪🧵 arxiv.org/abs/2504.02234
arxiv.org
LLM Social Simulations Are a Promising Research Method
Accurate and verifiable large language model (LLM) simulations of human research subjects promise an accessible data source for understanding human behavior and training new AI systems. However, resul...
3207
Reposted by @tiagogiraldo.bsky.social
Sung Kim @sungkim.bsky.social · 01/04/2025
The evolution of graph learning Google describes how graphs and graph learning have evolved since the advent of PageRank in 1996, highlighting key studies and research. research.google/blog/the-evo...
0161
Reposted by @tiagogiraldo.bsky.social
Science X / Phys.org @sciencex.bsky.social · 16/03/2025
Deep learning has identified seven genes linked to accelerated brain aging and 13 existing drugs that may target these genes, offering potential strategies to slow brain aging. doi.org/g88kbz
medicalxpress.com
Deep learning uncovers gene targets and potential drugs to slow brain aging
In a new study reported in Science Advances, scientists analyzed MRI data stored at the UK Biobank and identified seven genes responsible for fast biological brain aging and 13 existing drugs that can target those genes.
122
Reposted by @tiagogiraldo.bsky.social
Gradient Flow @gradientflow.com · 16/03/2025
✅✅ This latest collection of GUIDES and CHEAT SHEETS covers everything from Manus to Agents to Google's Gemma 3 and the evolving regulatory landscape across major economies #AI #LLM gradientflow.substack.com/p/ai-this-we...
gradientflow.substack.com
AI This Week: New Agents, Open Models, and the Race for Productivity
Subscribe • Previous Issues
011
Reposted by @tiagogiraldo.bsky.social
Maike Osborne @maosbot.bsky.social · 16/03/2025
Everything they've recorded is great, but this one is probably my favourite. When I first heard it, I literally laughed out loud youtu.be/-kz_CrB0Vf0?...
youtu.be
Lambrini Girls - Company Culture (Official Video)
YouTube video by Lambrini Girls
2142
Reposted by @tiagogiraldo.bsky.social
Kuan Hoong @kuanhoong.bsky.social · 12/03/2025
Gemma 3 is a collection of lightweight, state-of-the-art open models built from the same research and technology that powers Gemini 2.0 models. Gemma 3 comes in a range of sizes (1B, 4B, 12B and 27B) blog.google/technology/d...
blog.google
Introducing Gemma 3: The most capable model you can run on a single GPU or TPU
Today, we're introducing Gemma 3, our most capable, portable and responsible open model yet.
1152
Reposted by @tiagogiraldo.bsky.social
Tiago Peixoto @tiago.skewed.de · 12/03/2025
🚨 New paper on ArXiv: “Uncertainty quantification and posterior sampling for network reconstruction” TL;DR; We present an efficient method to sample the entire ensemble of possible network reconstructions that are compatible with an indirect observation, e.g. a dynamics. Short thread: 1/N
1016553
Reposted by @tiagogiraldo.bsky.social
Science X / Phys.org @sciencex.bsky.social · 12/03/2025
The James Webb Space Telescope has observed that most galaxies in the deep universe rotate in the same direction, challenging existing cosmological theories and suggesting new avenues for exploration. doi.org/g874s7
phys.org
Puzzling observation by JWST: Galaxies in the deep universe rotate in the same direction
In just over three years since its launch, NASA's James Webb Space Telescope (JWST) has generated significant and unprecedented insights into the far reaches of space, and a new study by a Kansas State University researcher provides one of the simplest and most puzzling observations of the deep universe yet.
042
Reposted by @tiagogiraldo.bsky.social
TechCrunch @techcrunch.com · 12/03/2025
Wolf Games, backed by ‘Law & Order’ creator, uses AI to create murder mystery games
tcrn.ch
Wolf Games, backed by ‘Law & Order’ creator, uses AI to create murder mystery games
Elliot Wolf, the executive producer and son of “Law & Order” creator Dick Wolf, is entering a new venture aimed at engaging true crime fans.  He, along with co-founders Andrew Adashek (CEO) and Noah Rosenberg (CTO), are developing Wolf Games, a new…
4155
Reposted by @tiagogiraldo.bsky.social
Sung Kim @sungkim.bsky.social · 12/03/2025
Mixture of Experts (MoE)! Built for interpretability, they create sparse, powerful models. No more post-training hacks (Bye-bye sparse auto-encoders) Paper: Mixture of Experts Made Intrinsically Interpretable ( arxiv.org/abs/2503.07639 )
0261
Reposted by @tiagogiraldo.bsky.social
Derek Law @gisbandit.bsky.social · 12/03/2025
Designing Effective #Dashboards tinyurl.com/49rw3ksj #ArcGISDashboards #dataviz #bestpractices #GIS #esri #arcgis #GISchat #geospatial #ArcGISApps #TheScienceOfWhere #geosky
0132
Reposted by @tiagogiraldo.bsky.social
Gradient Flow @gradientflow.com · 10/03/2025
The Rapid Rise of Generative AI 🆕 National Survey Reveals Adoption Rates Exceeding Early Internet Growth gradientflow.com/genai-adopti... #AI #LLM #GenerativeAI #GenAI
gradientflow.com
The State of Generative AI in America: First National Study Reveals Adoption Patterns and Productivity Gains - Gradient Flow
National Survey Maps Generative AI’s Integration into Daily Tasks. In a recent research study to measure the real-world impact of generative AI, economists Alexander Bick, Adam Blandin, and David J. D...
011
Reposted by @tiagogiraldo.bsky.social
Science X / Phys.org @sciencex.bsky.social · 10/03/2025
A new AI tool, PhyloFrame, addresses ancestral bias in genetic data, enhancing precision medicine by improving disease prediction and treatment across diverse populations. doi.org/g87p49
medicalxpress.com
Computer engineers develop AI tool to account for ancestral diversity in genetic data
University of Florida researchers are addressing a critical gap in medical genetic research—ensuring it better represents and benefits people of all backgrounds.
033
Reposted by @tiagogiraldo.bsky.social
British Antarctic Survey 🐧 @bas.ac.uk · 10/03/2025
Surviving AND thriving? This newly-discovered limpet from the Arctic is an icon for our times. ⏬ Good under pressure (3,053m deep) 🧊 Happy in freezing waters ❤️‍🔥 ALSO fine on 300°C hydrothermal vents 🧪 Fine with acidic water dissolving its shell... 💗 ...because it grows new shell from inside! 1/2
A photo of the limpet on a black background - photographed side on. It has a short cone-shaped shell with an open bottom, where you can see it's fleshy parts inside. The shell is marked by rows of ripple patterns, which are different areas of growth.
29431
Reposted by @tiagogiraldo.bsky.social
Quanta Magazine @quantamagazine.org · 10/03/2025
In a chaotic system, no detail is small enough to ignore. Adjusting the position of a butterfly in Brazil by a millimeter, in one infamous metaphor, could mean the difference between a typhoon striking Tokyo and a tornado tearing through Tennessee. buff.ly/SKLsRk0
0184
Reposted by @tiagogiraldo.bsky.social
Qiusheng Wu @giswqs.bsky.social · 10/03/2025
🎉 The recording of the first session of our Open Science Workshop series is now live! 📺 Watch it on YouTube: www.youtube.com/live/sh5iPQR... 🔍 Don’t miss our next workshop on Open Data happening March 24! Stay updated here: tiny.utk.edu/openscience #opencescience #coding #python #jupyter
youtube.com
Open Science Workshop 1: Open Code
YouTube video by Open Geospatial Solutions
095
Reposted by @tiagogiraldo.bsky.social
Tom Silver @tomssilver.bsky.social · 02/03/2025
This week's #PaperILike is "Model Predictive Control and Reinforcement Learning: A Unified Framework Based on Dynamic Programming" (Bertsekas 2024). If you know 1 of {RL, controls} and want to understand the other, this is a good starting point. PDF: arxiv.org/abs/2406.00592
arxiv.org
Model Predictive Control and Reinforcement Learning: A Unified Framework Based on Dynamic Programming
In this paper we describe a new conceptual framework that connects approximate Dynamic Programming (DP), Model Predictive Control (MPC), and Reinforcement Learning (RL). This framework centers around ...
0438
Reposted by @tiagogiraldo.bsky.social
Gradient Flow @gradientflow.com · 02/03/2025
Decoding Inference Scaling: The Secret to Better AI Performance 🎯 #AI #LLM → gradientflow.substack.com/p/boost-ai-p...
gradientflow.substack.com
Boost AI Performance: Understanding Inference Scaling
Subscribe • Previous Issues
011
Reposted by @tiagogiraldo.bsky.social
Simon Willison @simonwillison.net · 02/03/2025
Here are my favorite snippets from my recent appearance on the Accessibility + Generwtive AI podcast, including notes on using LLMs to help with alt text and the ethics of building accessibility tools on top of inherently unreliable technology simonwillison.net/2025/Mar/2/a...
simonwillison.net
Notes from my Accessibility and Gen AI podcast appearence
I was a guest on the most recent episode of the Accessibility + Gen AI Podcast, hosted by Eamon McErlean and Joe Devon. We had a really fun, wide-ranging conversation …
2473
Reposted by @tiagogiraldo.bsky.social
Norman @noaasm.bsky.social · 02/03/2025
Close to 12,000 valid secrets that include API keys and passwords have been found in the Common Crawl dataset used for training multiple artificial intelligence models. www.bleepingcomputer.com/news/securit...
bleepingcomputer.com
Nearly 12,000 API keys and passwords found in AI training dataset
Close to 12,000 valid secrets that include API keys and passwords have been found in the Common Crawl dataset used for training multiple artificial intelligence models.
0163
Reposted by @tiagogiraldo.bsky.social
Ramon Astudillo @ramon-astudillo.bsky.social · 09/02/2025
Reading another paper arguing little data suffices to extract the skills already developed in pre-training. LIMO arxiv.org/abs/2502.03387. First sip but, bit self-aggrandizing? Scrolled to see model (so excuse if I missed something, will report later) ... Its a Qwen-32b-instruct ... AIME score? 16.5
arxiv.org
LIMO: Less is More for Reasoning
We present a fundamental discovery that challenges our understanding of how complex reasoning emerges in large language models. While conventional wisdom suggests that sophisticated reasoning tasks de...
111
Reposted by @tiagogiraldo.bsky.social
Santiago Viquez @santiviquez.com · 09/02/2025
It's happening! Join us next week to ask Sebastian Raschka anything! 📅 Date: February 11th ⏰ Time: 10:00 AM – 11:00 AM EST 📍 Register: lu.ma/evqa4rct
011
Reposted by @tiagogiraldo.bsky.social
Valeriy M., PhD, MBA, CQF @predict-addict.bsky.social · 09/02/2025
**Big Names in Finance Are Betting Big on Conformal Prediction Innovation** The financial industry is undergoing a quiet revolution, and at the heart of it lies conformal prediction—a groundbreaking framework for uncertainty quantification in machine learning.
122
Reposted by @tiagogiraldo.bsky.social
Sung Kim @sungkim.bsky.social · 09/02/2025
Do Large Language Model Benchmarks Test Reliability? by @aleksmadry.bsky.social Do current LLMs perform simple tasks reliably? They find out the benchmarks are too noisy to pinpoint such lingering failures where over 50% of “model errors” are actually caused by benchmark errors.
2171
Reposted by @tiagogiraldo.bsky.social
TechCrunch @techcrunch.com · 09/02/2025
DeepSeek’s R1 reportedly ‘more vulnerable’ to jailbreaking than other AI models
tcrn.ch
DeepSeek’s R1 reportedly ‘more vulnerable’ to jailbreaking than other AI models
The latest model from DeepSeek, the Chinese AI company that’s shaken up Silicon Valley and Wall Street, can be manipulated to produce harmful content such as plans for a bioweapon attack and a campaign to promote self-harm among teens, according to The…
105221
Reposted by @tiagogiraldo.bsky.social
TechCrunch @techcrunch.com · 09/02/2025
OpenAI CEO Sam Altman admits that AI’s benefits may not be widely distributed
tcrn.ch
OpenAI CEO Sam Altman admits that AI’s benefits may not be widely distributed
In a new essay on his personal blog, OpenAI CEO Sam Altman said the company is open to a “compute budget,” among other “strange-sounding” ideas, to “enable everyone on Earth to use a lot of AI” and ensure the benefits of the technology are widely…
2211141
tiagogiraldo.bsky.social @tiagogiraldo.bsky.social · 08/02/2025
Para @nataliavartan.bsky.social, creo que te puede gustar ⬇️ "Terence Tao on how we measure the cosmos | Part 1" www.youtube.com/watch?v=YdOX...
youtube.com
Terence Tao on how we measure the cosmos | Part 1
YouTube video by 3Blue1Brown
120
Reposted by @tiagogiraldo.bsky.social
Gisela Baños @gisbanos.bsky.social · 08/02/2025
Recorre el camino de Mark Watney, el protagonista de #ElMarciano, de Andy Weir, por la superficie de Marte gracias a la NASA. trek.nasa.gov/mars/index.h...
2187
tiagogiraldo.bsky.social @tiagogiraldo.bsky.social · 08/02/2025
Karpathy's new video ⬇️ "Deep Dive into LLMs like ChatGPT" www.youtube.com/watch?v=7xTG...
youtube.com
Deep Dive into LLMs like ChatGPT
YouTube video by Andrej Karpathy
010
Reposted by @tiagogiraldo.bsky.social
Ethan Mollick @emollick.bsky.social · 07/02/2025
This paper is wild - a Stanford team shows the simplest way to make an open LLM into a reasoning model They used just 1,000 carefully curated reasoning examples & a trick where if the model tries to stop thinking, they append "Wait" to force it to continue. Near o1 at math. arxiv.org/pdf/2501.19393
721333
Reposted by @tiagogiraldo.bsky.social
Mark Riedl @markriedl.bsky.social · 07/02/2025
Face meet palm
1111
Reposted by @tiagogiraldo.bsky.social
Science X / Phys.org @sciencex.bsky.social · 07/02/2025
Quantum computers simulate particle creation in an expanding universe, aligning with theoretical predictions. This approach offers insights into early universe phenomena and advances cosmological research. doi.org/g8383g
phys.org
Simulating particle creation in an expanding universe using quantum computers
A new study published in Scientific Reports simulates particle creation in an expanding universe using IBM quantum computers, demonstrating the digital quantum simulation of quantum field theory for curved spacetime (QFTCS).
001
Reposted by @tiagogiraldo.bsky.social
Sebastian Raschka (rasbt) @rasbt.bsky.social · 07/02/2025
Just read the s1: Simple Test-Time Scaling paper. Super interesting approach to improving reasoning models! TL;DR: 1. SFT on 1k curated examples w/ reasoning traces. 2. Control response length w/ budget forcing: "Wait" tokens → longer reasoning/self-correction. "Final Answer:" → enforce stopping.
2376
Reposted by @tiagogiraldo.bsky.social
Gradient Flow @gradientflow.com · 06/02/2025
Transform Photos into Life-Like Videos 🌟 This AI Can Make Your Photos Sing and Dance ✨ OmniHuman is a 🆕 Diffusion Transformer-based framework from Bytedance omnihuman-lab.github.io
omnihuman-lab.github.io
OmniHuman-1 Project
OmniHuman-1: Rethinking the Scaling-Up of One-Stage Conditioned Human Animation Models
001
Reposted by @tiagogiraldo.bsky.social
Gradient Flow @gradientflow.com · 06/02/2025
Breaking Down AI Agents 🎯 10 Critical Aspects of AI Agents That Will Shape Their Future gradientflow.substack.com/p/10-things-...
gradientflow.substack.com
10 Things to Know About the State of AI Agents
Subscribe • Previous Issues
011
Reposted by @tiagogiraldo.bsky.social
Gradient Flow @gradientflow.com · 05/02/2025
Large Language Models and Foundation Model Futures: Key Trends for 2025 and Beyond #AI #LLM #LLMs gradientflow.com/foundation-m...
gradientflow.com
Foundation Models: What's Next for 2025 and Beyond - Gradient Flow
Recent releases from Google (Gemini 2.0 Flash), OpenAI (o1 & o3), and particularly DeepSeek (V3 & R1) have underscored the rapid pace of innovation in the foundation model space. This prompted me to c...
041
Reposted by @tiagogiraldo.bsky.social
Ethan Mollick @emollick.bsky.social · 05/02/2025
One thing academics should take away from Deep Research is that a substantial number of your readers in the future will likely be AI agents. Is your paper available in an open repository? Are any charts and graphs described well in the text? Is it well-titled? Probably worth considering these…
2869
Reposted by @tiagogiraldo.bsky.social
TechCrunch @techcrunch.com · 04/02/2025
Hugging Face researchers aim to build an ‘open’ version of OpenAI’s deep research tool
tcrn.ch
Hugging Face researchers aim to build an ‘open’ version of OpenAI’s deep research tool
A group of developers at AI dev platform Hugging Face, including Thomas Wolf, the company’s co-founder and chief scientist, say they’ve built an “open” version of OpenAI’s deep research tool. Deep research, which OpenAI unveiled during an event Sunday,…
35610
Reposted by @tiagogiraldo.bsky.social
freeCodeCamp.org @freecodecamp.bsky.social · 04/02/2025
LangChain has been a go-to choice for creating AI applications for a while now. But maybe you want to explore other options that are more flexible, simpler, and less expensive. Here, Maham walks you through LangChain alternatives for building AI and agentic workflows. buff.ly/4hmh3HT
062
Reposted by @tiagogiraldo.bsky.social
Qiusheng Wu @giswqs.bsky.social · 04/02/2025
🚀 I will teaching a 90-min workshop at 9AM EST tomorrow (Feb. 5). It is free and open to the public. There is still time to register. Title: Mastering remote sensing image segmentation with AI: A hands-on workshop with the Segment Anything Model Register: aiforgood.itu.int/event/master...
071
Reposted by @tiagogiraldo.bsky.social
midudev · Miguel Ángel Durán @midu.dev · 04/02/2025
No sé por qué la gente se queja de DeepSeek si el código que crea es brutal:
Código de algún lenguaje de programación en chino
2273
Reposted by @tiagogiraldo.bsky.social
Sung Kim @sungkim.bsky.social · 04/02/2025
Cross-Entropy Loss is NOT What You Need! They introduce harmonic loss as an alternative to the standard CE loss for training neural networks and LLMs! Harmonic loss achieves 🛠️significantly better interpretability, ⚡faster convergence, and ⏳less grokking!
6413