ruggsea
@ruggsea.bsky.social
AI/NLP research at @uni-graz.at, (Data) Journalism for scomodo.org.
Before: AI policy & Data science @_interface_eu, geopolitical analysis for @Geopoliticainfo
Before: AI policy & Data science @_interface_eu, geopolitical analysis for @Geopoliticainfo
The more I spend time in Austria, the more this feels true
November 4, 2025 at 4:58 PM
The more I spend time in Austria, the more this feels true
We are hiring a Senior Scientist! If questions like "how much of Common Crawl can we actually store and process on our infrastructure?" get you excited, you would prob love this job. Come do research at fun scales with us
My lab is looking for a Senior Scientist (= PostDoc with option of permanency)!
We are looking for someone interested in doing cutting-edge computational social science + helping us with data & software engineering 🤓.
See job ad for details jobs.uni-graz.at/en/jobs/7d14...
We are looking for someone interested in doing cutting-edge computational social science + helping us with data & software engineering 🤓.
See job ad for details jobs.uni-graz.at/en/jobs/7d14...
Universität Graz
jobs.uni-graz.at
October 28, 2025 at 6:07 PM
We are hiring a Senior Scientist! If questions like "how much of Common Crawl can we actually store and process on our infrastructure?" get you excited, you would prob love this job. Come do research at fun scales with us
A thing I've been working on for the past year: an LLM benchmark on the dreaded Italian medicine faculty entry exam. I will present it in two weeks at @ailc-nlp.bsky.social Clic-it in Cagliari!
Ruggero Marino Lazzaroni, Alessandro Angioi, Michelangelo Puliga, Davide Sanna, Roberto Marras
MedBench-IT: A Comprehensive Benchmark for Evaluating Large Language Models on Italian Medical Entrance Examinations
https://arxiv.org/abs/2509.07135
MedBench-IT: A Comprehensive Benchmark for Evaluating Large Language Models on Italian Medical Entrance Examinations
https://arxiv.org/abs/2509.07135
September 10, 2025 at 12:47 PM
A thing I've been working on for the past year: an LLM benchmark on the dreaded Italian medicine faculty entry exam. I will present it in two weeks at @ailc-nlp.bsky.social Clic-it in Cagliari!
from this awesome blogpost: snats.xyz/pages/articl...
September 9, 2025 at 9:27 AM
from this awesome blogpost: snats.xyz/pages/articl...
Life is going from one cool spot to train AI models to another
In this fancy Berlin library you can listen to vinyls while you do it!
In this fancy Berlin library you can listen to vinyls while you do it!
July 30, 2025 at 9:26 AM
Life is going from one cool spot to train AI models to another
In this fancy Berlin library you can listen to vinyls while you do it!
In this fancy Berlin library you can listen to vinyls while you do it!
Me looking very dumb while pointing at things at two academic events:
1. Pointing at a logits inside @repligate.bsky.social's loom for the "Braive New World" conference at @uni-graz.at
2. Pointing at my poster at @ic2s2.bsky.social last week on improving LLM agentic natural conversation synthesis
1. Pointing at a logits inside @repligate.bsky.social's loom for the "Braive New World" conference at @uni-graz.at
2. Pointing at my poster at @ic2s2.bsky.social last week on improving LLM agentic natural conversation synthesis
July 28, 2025 at 1:56 PM
Me looking very dumb while pointing at things at two academic events:
1. Pointing at a logits inside @repligate.bsky.social's loom for the "Braive New World" conference at @uni-graz.at
2. Pointing at my poster at @ic2s2.bsky.social last week on improving LLM agentic natural conversation synthesis
1. Pointing at a logits inside @repligate.bsky.social's loom for the "Braive New World" conference at @uni-graz.at
2. Pointing at my poster at @ic2s2.bsky.social last week on improving LLM agentic natural conversation synthesis
Only prisoners have time to read, and if you want to engage in a twenty-year long research project funded by the state, you will have to kill someone.
Sorry for the Fisher posting but it's so good
Sorry for the Fisher posting but it's so good
July 27, 2025 at 1:36 PM
Only prisoners have time to read, and if you want to engage in a twenty-year long research project funded by the state, you will have to kill someone.
Sorry for the Fisher posting but it's so good
Sorry for the Fisher posting but it's so good
I look really bad/funny in this picture but I am glad I was given the opportunity to talk about LLM research to an audience of cool researchers!
Fun fact: I also held a hands-on session that involved playing around with Bluesky data!
Fun fact: I also held a hands-on session that involved playing around with Bluesky data!
Scientists know it's happening: #LLMs like #ChatGPT are quietly transforming #academia —helping write papers, draft grants & process data. CSH's @lespin.bsky.social organized a 3-day workshop to explore their ethical, practical use in research—from writing to coding to data annotation.
July 10, 2025 at 2:39 PM
I look really bad/funny in this picture but I am glad I was given the opportunity to talk about LLM research to an audience of cool researchers!
Fun fact: I also held a hands-on session that involved playing around with Bluesky data!
Fun fact: I also held a hands-on session that involved playing around with Bluesky data!
Karpathy on training Neural Networks: you should go slowly and be paranoid
me, while vibecoding torch code: what if i just increase the paranoid part?
me, while vibecoding torch code: what if i just increase the paranoid part?
June 9, 2025 at 2:27 PM
Karpathy on training Neural Networks: you should go slowly and be paranoid
me, while vibecoding torch code: what if i just increase the paranoid part?
me, while vibecoding torch code: what if i just increase the paranoid part?
how not to name your ai agents
June 3, 2025 at 7:19 PM
how not to name your ai agents
genz semantic embeddings engineering from my collegue
May 27, 2025 at 10:48 AM
genz semantic embeddings engineering from my collegue
backpropagation was inspired by freud
May 21, 2025 at 10:53 AM
backpropagation was inspired by freud
“How about we pull over for a bit and get some rest?" - GPT4, when it's their turn to drive
May 20, 2025 at 9:23 PM
“How about we pull over for a bit and get some rest?" - GPT4, when it's their turn to drive
I wish all textbooks were written like this one from @jurafsky.bsky.social
May 18, 2025 at 11:14 PM
I wish all textbooks were written like this one from @jurafsky.bsky.social
was true in 2012 and it is true now
May 16, 2025 at 9:56 AM
was true in 2012 and it is true now
Stumbled upon new ammunition for my personal struggle against RHLF: the tendency of models trained this way to give correct sounding answers make us believe in wrong things more
arxiv.org/abs/2409.12822
arxiv.org/abs/2409.12822
Language Models Learn to Mislead Humans via RLHF
Language models (LMs) can produce errors that are hard to detect for humans, especially when the task is complex. RLHF, the most popular post-training method, may exacerbate this problem: to achieve h...
arxiv.org
May 15, 2025 at 1:03 AM
Stumbled upon new ammunition for my personal struggle against RHLF: the tendency of models trained this way to give correct sounding answers make us believe in wrong things more
arxiv.org/abs/2409.12822
arxiv.org/abs/2409.12822
Looks important for my rec algorithm people: hard confirmation that Google uses essentially hybrid rankings for search: page rank, bert embeddings and a secret ingredient (user data)
New Google/DOJ docs realised and to my knowledge the first time we've had a primary source specifically saying Google uses Chrome data in a popularity signal for ranking. 👀
May 13, 2025 at 9:07 PM
Looks important for my rec algorithm people: hard confirmation that Google uses essentially hybrid rankings for search: page rank, bert embeddings and a secret ingredient (user data)
All of this mentions of Zero make me think about Slothrop
May 8, 2025 at 10:31 AM
All of this mentions of Zero make me think about Slothrop
Reposting from the other site to spread it here: apparently, thinking that RLHF irons out creativity in LLMs is now corroborated by this paper arxiv.org/pdf/2505.00047
May 6, 2025 at 11:45 PM
Reposting from the other site to spread it here: apparently, thinking that RLHF irons out creativity in LLMs is now corroborated by this paper arxiv.org/pdf/2505.00047
Protestant-Catholic German beefs are my favorite kind of beefs
April 9, 2025 at 1:23 PM
Protestant-Catholic German beefs are my favorite kind of beefs
At Weizenbaum today, talking about slightly more complicated chatbots!
April 3, 2025 at 12:04 PM
At Weizenbaum today, talking about slightly more complicated chatbots!
A friend and colleague at my lab (the CS² under @janalasser.bsky.social ) is organizing a super nice school on democracy and social media here in Graz!
If you're interested, check it out and share! Link: sicss.io/2025/graz/
If you're interested, check it out and share! Link: sicss.io/2025/graz/
March 27, 2025 at 2:46 PM
A friend and colleague at my lab (the CS² under @janalasser.bsky.social ) is organizing a super nice school on democracy and social media here in Graz!
If you're interested, check it out and share! Link: sicss.io/2025/graz/
If you're interested, check it out and share! Link: sicss.io/2025/graz/
Claude, my friend, are you ok?
March 24, 2025 at 3:56 PM
Claude, my friend, are you ok?
Has anybody ever tried GRPO training with annealing wrt the generations temperature?
I feel like that could be a way to explore more of the search space at the beginning of the training, to then be less "creative" once the model is hopefully already on the right track
I feel like that could be a way to explore more of the search space at the beginning of the training, to then be less "creative" once the model is hopefully already on the right track
March 20, 2025 at 3:22 PM
Has anybody ever tried GRPO training with annealing wrt the generations temperature?
I feel like that could be a way to explore more of the search space at the beginning of the training, to then be less "creative" once the model is hopefully already on the right track
I feel like that could be a way to explore more of the search space at the beginning of the training, to then be less "creative" once the model is hopefully already on the right track
