Evren @ecrtmt

@elonmusk What do @elonmusk and Soros have in common? Both are billionaires trying to influence global politics.

"Move 37" is the word-of-day - it's when an AI, trained via the trial-and-error process of reinforcement learning, discovers actions that are new, surprising, and secretly brilliant even to expert humans. It is a magical, just slightly unnerving, emergent phenomenon only achievable by large-scale reinforcement learning. You can't get there by expert imitation. It's when AlphaGo played move 37 in Game 2 against Lee Sedol, a weird move that was estimated to only have 1 in 10,000 chance to be played by a human, but one that was creative and brilliant in retrospect, leading to a win in that game. We've seen Move 37 in a closed, game-like environment like Go, but with the latest crop of "thinking" LLM models (e.g. OpenAI-o1, DeepSeek-R1, Gemini 2.0 Flash Thinking), we are seeing the first very early glimmers of things like it in open world domains. The models discover, in the process of trying to solve many diverse math/code/etc. problems, strategies that resemble the internal monologue of humans, which are very hard (/impossible) to directly program into the models. I call these "cognitive strategies" - things like approaching a problem from different angles, trying out different ideas, finding analogies, backtracking, re-examining, etc. Weird as it sounds, it's plausible that LLMs can discover better ways of thinking, of solving problems, of connecting ideas across disciplines, and do so in a way we will find surprising, puzzling, but creative and brilliant in retrospect. It could get plenty weirder too - it's plausible (even likely, if it's done well) that the optimization invents its own language that is inscrutable to us, but that is more efficient or effective at problem solving. The weirdness of reinforcement learning is in principle unbounded. I don't think we've seen equivalents of Move 37 yet. I don't know what it will look like. I think we're still quite early and that there is a lot of work ahead, both engineering and research. But the technology feels on track to find them. youtube.com/watch?v=HT-UZk…

When humanity does create AGI, it will be named Untitled14.ipynb

showing up to the airport at 6am for your €21.99 flight

Ramin Setoodeh @RaminSetoodeh

2 years ago

Cillian Murphy #GoldenGlobes

47 344 4K 4.5M 552

@TheOneLanceB That should be the advantage of ICT concepts - they work on FOREX, futures or stocks. That's what I heard.

@casedeepoca It is the oldest mosque from Romania.

@predict_addict I was using the default scikit-learn methods. Calibration seems to be dependent on model performance and dataset imbalance. Calibration seems easier when the imbalance is not that high and the model is able to overcome underfitting.

@robbiewilliams was the KING

Göbekli Tepe, located in the southeastern Anatolia region of Turkey near the city of Urfa (also known as Şanlıurfa), is considered the oldest known temple complex in the world. It is an archaeological site that has captivated researchers and historians with its remarkable age and significance. Discovered in the 1960s by the archaeologist Klaus Schmidt, Göbekli Tepe dates back approximately 11,000 years, making it older than Stonehenge in England and the Great Pyramids of Egypt. The site was built and used by ancient hunter-gatherer societies during the Pre-Pottery Neolithic period, challenging the conventional belief that monumental architecture was a product of settled agricultural communities. Göbekli Tepe comprises multiple stone pillars arranged in circular structures, which are thought to have served as ritualistic gathering places. These pillars, some of which reach up to 16 feet (5 meters) in height and weigh several tons, are adorned with intricate carvings of various animals, including lions, snakes, foxes, and birds. What makes Göbekli Tepe particularly fascinating is the complexity and sophistication of its construction. The massive stones used to create the pillars were transported from quarries several miles away, and their shaping and assembly would have required a high level of coordination and organization. This suggests that the hunter-gatherer societies that built Göbekli Tepe possessed a degree of social organization and communal effort not typically associated with their era. The purpose of Göbekli Tepe remains a subject of speculation and debate among archaeologists. Some believe it was primarily a religious or ceremonial site, where rituals and gatherings took place. The carvings of animals suggest a connection to ancient belief systems and potentially represent the worship of animal spirits or deities. Others theorize that it may have served as a meeting place for different groups or as a center for trading and exchange of goods. The discovery of Göbekli Tepe has had a significant impact on our understanding of human history and the development of civilization. It challenges the traditional narrative that agriculture was the driving force behind the rise of complex societies and suggests that religious or social factors may have played a more prominent role. The site provides valuable insights into the cultural and technological achievements of early humans, highlighting their capacity for architectural innovation and symbolic expression. Since its discovery, Göbekli Tepe has been meticulously excavated and studied by archaeologists, shedding new light on our ancient past. It has been recognized as a UNESCO World Heritage site, attracting visitors from around the world who come to marvel at this extraordinary testament to human ingenuity and the dawn of civilization.

@andreisavu Can't beat ChatGPT, but for evaluating subsets of a dataset for a given metric, I found TFMA useful tensorflow.org/tfx/model_anal…

The AI Trinity

Overheard: “People who know nothing about machine learning are now paradoxically advantaged in LLMs because they don’t immediately reach for overly sophisticated ideas and spend a lot more time hacking prompts” When hacking prompts feels below your dignity but it works :’|

@elonmusk @mattyglesias It's funny that @elonmusk is always in the position of choosing between 2 impossible choices: firing 80% of Twitter employees vs going bankrupt, have Twitter throttled in its entirety or limit access to some tweets.

Optimists vs Pessimists

Geoffrey Hinton leaving Google, so he can freely speak about the dangers of the technology he worked on, reminds me of the following quote from Robert Oppenheimer: “Now I am become Death, the destroyer of worlds” youtu.be/lb13ynu3Iac

A tasty snooker from Ronnie!

@oanaolt Let's go Constanta!

#llmops #mlops #whatsnext

@elonmusk Search box and tags for bookmarks.