If Europeans wouldn’t sleep on this opportunity, we could massively improve our science recruiting. Related scientific paper of what getting the fired scientists from an autocracy can do to your science system: https://www.aeaweb.org/articles?id=10.1257/aer.104.10.3222

That is something that some tech savy Lemmy users could already easily do. I repost stuff from all over the web. But some systematic preservation of good old subreddits aught to be automated.

Didn’t deepseek solve some of the data wall problems by creating good chain of thought data with an intermediate RL model. That approach should work with the tried and tested scaling laws just using much more compute.

doesn’t deepseek work on that though with their janus models?

Important context and a good decision

4chan at least had a consistent brand of being the anti-social network and being full of Nazis, weirdos, pedophiles and people who are just anti-social for the lulz. You couldn’t ruin 4chan.

Twitter’s image was being the “internet town-square for serious thinkers” with politicians, scientists, journalists and a small but good measure of standard shitposters. Loosing that brand diminishes it’s value massively. Unfortunately neither Bluesky nor Mastodon was able to catch that clientele yet.

It’s the famous “As long as your not Google, Amazon or Apple” licence.

The key story for Christians is that a man was unjustly executed by religious authorities and the state even though he was without sin. This unjust sacrifice atones for all of humanities sin. So kinda hard to justify execution from a Christian perspective if you actually believe that stuff.

So this school was built on an ancient Pleistocene burial ground. I know that trope well enough to know what happened next

There are tons of statistical methods to get reasonable conclusions without an RCT. Some things can not be detected with an RCT, because the experiment is just impossible to run, so sometimes you need methods to do causal identification with observable data. Here you do not even need causal identification methods for observational data. You just need to do some descriptive statistics for a large group of people well to find interesting patterns. Whether this aging pattern in the mid-40s is causal it coincidental is not important at first. The pattern itself is interesting.

I was on a holiday in the Cinque Terre in Italy with my wife a few years ago. Because of a rainy day we decided to take a train to Genua and visit some museums. At the maritime museum I randomly met an Italian coworker/coauthor from my research institute in Germany, who was visiting his family in his hometown with his wife.

For a user without much technical experience using a ready-made gui like Jan.ai with automatic model download and ability to run models with the ggml library on consumer grade hardware like mac M-series chips or cheap GPUs by either Nvidia or AMD is probably a good start.

For a little bit more technically proficient users Ollama is probably a great choice to start to host your own OpenAI-like API for local models. I mostly run gemma2 or small llama 3.1 like models with that.

The market will segment away from the current tech anyway. CATL Sodium-ion with comparatively low densities but also extremely low prices per kWh will likely win the low-end market and the market for stationary solutions. This is just due to the much lower resource costs. The high-end will be up for things like this battery by Samsung (or other comparable pilot products). The current technology will likely be in a weird middle spot.

Depends on what you do with it. Synthetic data seems to be really powerful if it’s human controlled and well built. Stuff like tiny stories (simple llm-generated stories that only use the complexity of a 3-year olds vocabulary) can be used to make tiny language models produce sensible English output. My favourite newer example is the base data for AlphaProof (llm-generated translations of proofs in Math-Papers to the proof-validation system LEAN) to teach an LLM the basic structure of Mathematics proofs. The validation in LEAN itself can be used to only keep high-quality (i.e. correct) proofs. Since AlphaProof is basically a reinforcement learning routine that uses an llm to generate good ideas for proof steps to reduce the size of the space of proof steps, applying it yields new correct proofs that can be used to further improve its internal training data.

Edward Teller is just the kind of scientist you need to build civil engineering projects out of doomsday devices.

The people over at NCD must be getting raging hardons just from seeing this.

Welcome to the Nether Realm, a place of wonder and mystery. Enjoy seamless connectivity with Vodafone. If you have a plan with a flat rate for minutes/SMS or data, it does not apply here.

Otherwise, the standard rate applies: outgoing calls to Vodafone numbers 50 ct/min, other numbers 75 ct/min, incoming calls 30 ct/min, SMS to Vodafone numbers 25 ct/SMS, others 40 ct/SMS, data usage 1 EUR/MB. Wi-Fi calls will be charged as international calls from your home country to the Nether Realm.

For more info, contact Vodafone customer service at +44123456789

Your Vodafone Team

Na SpaceX would just use his neuralink chip it to automate the team that keeps Musk distracted from messing with important things in the company with a simple AI

Driving in Paris is crazy as hell, and I did drive a lot in European cities. Although the worst I’ve ever been to has to be Bucharest. People drive hyper-crazy in Romania.

Na Ghent and Brugges are lovely. Brussels is kinda meh but not superbad.