Josip Krapac @josipk.bsky.social · 1hThe word "needless" is carrying a lot of weight here. I didn't know that sometimes there is a need to be abusive and cruel, towards their models and in general. 000
Josip Krapac @josipk.bsky.social · 09/10/2026Ja sam imao ponavljajući san da se vozim vlakom koji je spor, nema krov, lagano pada kiša, a na nemam kišobran. 020
Josip Krapac @josipk.bsky.social · 08/10/2026We are living in the dark ages. Yes, things have improved immensely in the last 200 years due to technology, but that improvement was not followed always by improvement in human compassion. Especially in the last 45-50 years. We can still stop it, but we're numbed. 000
Josip Krapac @josipk.bsky.social · 07/10/2026I understand. They took "the last mile missing problems", which are still important, but where a lot of work has already been done, and even there they didn't do great. 110
Josip Krapac @josipk.bsky.social · 07/10/2026Why it's boring? Because it's mechanical application of existing techniques rather than invention of new techniques, or something else? 100
Josip Krapac @josipk.bsky.social · 04/10/2026Can you give an example where low returns per year in the beginning led to success? I can think only of Amazon (maybe other retailers too). I don't see many parallels between Amazon and OpenAI or Anthropic, but maybe you do. Or maybe you had other, non-retail businesses in mind for comparison? 100
Josip Krapac @josipk.bsky.social · 17/09/2026"We don't need any regulation, we've got market forces." One could say the same for water safety: market will do it's job when people start feeling the consequences of bad quality. Market doesn't care about people. People (should) do. Assuming that the companies will self-regulate is a pipe dream. 011
Josip Krapac @josipk.bsky.social · 03/09/2026The question is if the town will stop dying because of that enterprise. I think people don't think so, at least in their interpretation of "dying". Otherwise they would be for it. Either that, or they are wrong: the town would stop dying, even in their interpretation of "dying". 100
Josip Krapac @josipk.bsky.social · 03/09/2026Often the impact of economic activity on the wellbeing of the community (noise, pollution, environmental impact) is negative. The question then is if the gains from economic activity can be used to offset such negative effects. In a lot of cases the negative impacts are shared, but gains are not. 140
Josip Krapac @josipk.bsky.social · 30/08/2026Slavs heard the Germans and concluded they were as good as mute. 010
Josip Krapac @josipk.bsky.social · 30/08/2026Going in the opposite direction, from an existing word to the name for a group of people, the word for "German" in the Croatian language (but also, more generally, in all Slavic languages) is "Nijemac." The word "nijem" means "mute" (unable to speak). 110
Josip Krapac @josipk.bsky.social · 29/08/2026A lot of the things that are happening in the world today, and that were always happening, can be decomposed into these two axes: 1. freedom & death vs. regulation & survival, 2. alone vs. together. Today we're just experiencing a shift towards more freedom and death alone. 000
Josip Krapac @josipk.bsky.social · 29/08/2026In the end he said: "French people like their freedom, usually the freedom to die." We spent the rest of the ride in the gondola up the glacier in silence. 100
Josip Krapac @josipk.bsky.social · 29/08/2026Freedom and death go together. A French mountain guide told me 15 years ago that every year they find some people in the glacier crevasses because they go to the glacier alone, don't tell anyone, fall into a crevasse, and die because there's no signal. 120
Josip Krapac @josipk.bsky.social · 29/08/2026I thought the point is you can have longer summer vacations because you don't have to go back because kids start school, but maybe you're right. 010
Josip Krapac @josipk.bsky.social · 27/08/2026Maybe laissez-faire approach will bring us there. For example, remove regulation about vaccinating people for measles, people get measles and die, people resist because they love "freedom to die from measles" for some time, the new generation says WTF and brings back vaccine mandates. 000
Josip Krapac @josipk.bsky.social · 27/08/2026On the other hand China is regulating impacts of AI, not AI itself, e.g. interpreting existing laws so you can't fire people because AI has replaced them. 100
Josip Krapac @josipk.bsky.social · 27/08/2026But in the US even the ones you mentioned are getting less and less regulated. In China the biotech research is much faster because there's much less regulations, even compared to the US. It's not easy to regulate well, but I agree that regulation is key to sustainable progress. 110
Josip Krapac @josipk.bsky.social · 27/08/2026That's not robust and long term is not sustainable, but if you don't believe that future will happen this is the only thing that makes sense to do. 000
Josip Krapac @josipk.bsky.social · 27/08/2026That doesn't lead us anywhere worth going. On one hand, it does breed a particular type of innovation, the one that cares about maximizing just one objective. 100
Josip Krapac @josipk.bsky.social · 27/08/2026That means that we get emergent hierarchies. They are natural, but I wouldn't say they are good, at least not in my opinion. Peterson says: don't resist it, don't "flatten" it, climb it instead. 200
Josip Krapac @josipk.bsky.social · 27/08/2026But why is that so hard to see? I suppose it's because "free" and "regulated" are opposites. They really are, but all that we've accomplished as species that makes us human was by constraining laissez-faire. Right now "free" (or rather, a particular flavor of "free") is reigning. 200
Josip Krapac @josipk.bsky.social · 25/08/2026"to do Melnikov analysis for design of new rotors" ROTFL! 000
Josip Krapac @josipk.bsky.social · 23/08/2026It's *air* humidity. If you fill everything with water there's no air, hence no air humidity to measure. You can't go over 100% air humidity by definition: the excess water just starts to condense because air can't hold it. 130
Josip Krapac @josipk.bsky.social · 23/08/2026Full saturation (100% humidity) means that the air can't hold more water vapor at that temperature. 120
Josip Krapac @josipk.bsky.social · 23/08/2026For the same set of prompts how good is the best open vs. best closed (good = completes the task with acceptable quality), how many output tokens are produced for each one, and what is the price per token. Difficult to understand what is happening without this data IMO. 100
Josip Krapac @josipk.bsky.social · 20/08/2026Pa da, uvijek možeš reći da je biljka na posudbi dok ne da mladice. Kakav lopov! 031
Josip Krapac @josipk.bsky.social · 20/08/2026With dyslexia it's always fun. Have read "dreck" instead of "deck". 000
Josip Krapac @josipk.bsky.social · 20/08/2026Moj otac bi sigurno rekao: "Lopov sigurno jesi, a jesi li spasitelj ćemo tek vidjeti". Uvijek pun mudrosti i motivacije. 010
Josip Krapac @josipk.bsky.social · 20/08/2026Why not both? Nešto obratno od Sir Olivera iz Alan Forda ("Manje džentlmen, više lopov") ti si "Više spasitelj, manje lopov". 210
Josip Krapac @josipk.bsky.social · 15/08/2026Open: you can run them on your own hardware, don't need to send your data to anyone, you can tune them for your use cases and purposes. Closed: none of these things. 100
Josip Krapac @josipk.bsky.social · 15/08/2026Chinese LLMs/agents are open weights, US ones are not (except Meta's and Google's, but they haven't matched the quality of Chinese ones, until recently, and Google had put more emphasis on its closed weights models). 110
Reposted by Josip KrapacIain Murray @imurray.bsky.social · 08/08/2026Sharing this postdoc opportunity at Imperial to develop and apply machine-learning and structural-bioinformatics methods for protein design, with focus on modern generative approaches such as protein diffusion models and flow-based models.imperial.ac.ukDescriptionPlease note that job descriptions are not exhaustive, and you may be asked to take on additional duties that align with the key responsibilities ment... 044
Reposted by Josip KrapacMile Sikic @msikic.bsky.social · 25/07/2026Preferred candidates: - Experience with deep learning models or theoretical machine learning. Not looking for computational biologists for these positions - We are looking exclusively for those who are ready to come to Singapore - Those interested in creating biotech startups are more than welcome 021
Reposted by Josip KrapacMile Sikic @msikic.bsky.social · 25/07/2026AI in biology must be experimentally verifiable, and that requires wet labs. @astar_gis is hiring pre- and postdocs with proven AI knowledge to build foundation models for RNA therapeutics: mRNA vaccines/drugs, gene editing, ribozymes and in vivo CAR-T. Singapore-based. DM me and please repost. 113
Josip Krapac @josipk.bsky.social · 15/07/2026"Soon" is probably different time span in different in three circles, but GOD KNOWS I'd like it to be same. Determined, of course, by World Cup, because that's the only even with clearly defined end date. 010
Josip Krapac @josipk.bsky.social · 24/06/2026Nice small computer science lab in Quimper, count me in! Did post-doc in Rennes and only ♥️ for BZH. 010
Josip Krapac @josipk.bsky.social · 24/06/2026Why he's doing this? What's he trying to accomplish? 000
Josip Krapac @josipk.bsky.social · 24/06/2026And latest crescendo: making life miserable for his best engineers by reviewing LLM generated code, installing software that tracks everything they do, and finally getting into betting, pardon, prediction markets. 100
Josip Krapac @josipk.bsky.social · 24/06/2026Has Mark Zuckerberg finally lost it? It started with metaverse obsession that was going on for a couple of years and a couple of billions of dollars. Then came mass hiring, followed by the first out of many mass layoffs. Than he brought Scale.ai guy to lead AI in Meta. 100
Josip Krapac @josipk.bsky.social · 21/06/2026We were promised Washington swamp will be drained. Looks like the opposite is happening. 010
Josip Krapac @josipk.bsky.social · 06/06/2026Check this in a year. It will be interesting to see where Denis and Yann will be, and what will be happening to US science, engineering and economy. I have no idea, but my money in on those two (but a bit more on Yann). 000