cma

Interestingly: "For logic proof, it had been thought that AI could handle all logic problems in the near future, hence logic problems of solving conjectures might not be so interesting in the future."

Someone could have a small private breakthrough tomorrow that gives sample learning efficiency of the brain, online learning, and consolidated memories.

It could be as simple as do open releases and publication at first to help recruit talent who want that or who want to make a name for themselves. learned from the likes of... OpenAI, Google, Meta, Emad

It's TBP's fastest growing brand, 50/50 joint venture with Carlson. Mostly sold online, not in stores, lending towards podcast promotion over seeing in a convenience store. I'm not talking about the whole nicotine pouch trend here, I'm just responding to the claim here that he's driving 0 sales, and before to the claim nicotine pouches aren't heavily marketed.

For mass TV advertising etc. you don't see it a ton because of pretty heavy restrictions (not as strict as tobacco which is completely banned).

Maybe so, OP mentioned they were very similar and that still seems to hold for stainless. How about the aggregates within? The composite makeup can result in different CoTE than the individual aggregates, I think that's one part of concrete sidewalks cracking though, sometimes near shaded/unshaded boundaries. Roman concrete supposedly had some self healing properties before the cracks grew, from the lime inclusions.

I was just responding to "The harness is what takes these random and hallucinogenic models and make them into something deterministic and useful."

You can compare Fable vs Sol vs Kimi in the same harness if you want too and there are meaningful big differences. I chose all Anthropic ones to be safe from the they were finetuned on different harnesses complaint that would be made from that comparison.

You can run the same harness on fable, opus, sonnet, and see a huge difference between them. It is true the harness is important, and openai has begun encryption its instructions to swarmed sub-agents instead of just encrypting the chain of thought, but the model is still important at this stage.

There's also stainless steel rebar

> Nasdaq 100 has always been marketed as a tech-forward index. It would be a bit ridiculous if they didn’t include the most value tech companies on the market.

What if they floated only .01%? What's the cutoff for it being ridiculous not to include?

One example would be go to college/graduate school to invest in skills or get a job now to earn wealth over the same time period.

They tested them on formal languages of different power, and saw where they could generalize beyond the training data. Transformers failed to generalize pretty early on at stack machines/brace matching.

It was a good bit older of a paper though, if I remember it's somewhat expected from the pure feed forward nature of them and limited circuit depth, where LSTMs have some recurrence.

> considering that any nicotine product very much does have a cancer risk.

Source? The carcinogens are from other parts of tobacco than nicotine. Pouches are newer, but nicotine gum and patches have been studied for years.

Lots of podcasts advertise them, I think Tucker Carlson, one of the biggest podcasts in the world, has his own brand or promotion deal. I don't know if he is popular with young adults or still has his fox news demographic.

It's much easier to control dose and taper off than cigarettes. Doesn't seem that much worse than something like coffee. I think it has some heart risks, but gets rid of the cancer risk of cigarettes and dipping.

Not exactly the spam example, but Ireland remained a net exporter of food during the potato famine. The good farmland went to cattle and things like that for export.

This deepmind paper showed big limitations on nested structures with transformers, LSTMs actually did better:

Neural Networks and the Chomsky Hierarchy

https://arxiv.org/abs/2207.02098

But he's also using AI for formally verified math and for ideas in solving math problems. The part about it being ok because it is a supplement just means ok that these aren't formally verified and may have bugs, and may also mean ok to not credit the AI for the paper as it is just a visual supplement and not the main work.

Compare Claude at the time to 4o's sycophantic era (when o4 updated in response to Grok), Claude was lightyears better.

One of the big things that enabled extended context to work was training techniques for extending LLMs with reasoning, part of the thing he was saying wouldn't work.

> Maybe the new tattoos are just like being racist or something, but that’s hard to do when your heart isn’t in it and they will eventually find some way to absorb that.

I think I'd rather read slop than edgelord.

> What exactly was he dead wrong about that is proven by any of this?

He said as you need more and more tokens models will fall apart because each additional token is a chance for a mistake and they will just exponentially fall apart. But in practice models have learned to identify and self-correct mistakes and if you look at the graphs more inference reasoning tokens almost always give far better accuracy.

The scaling with reasoning models is more and more with things like verifiable rewards (coding and math), in line with bitter lesson and also Sutton invented lots of modern RL.

Why would auto mode turn off thinking?

Were your question in math areas related to ML? They also restrict model development and research pretty heavily.

Are you running unquantized GLM-5.2 and getting in loops or quantized?

By bunch up in a ball I just mean assume the fetal position when powered off, not wheels. They wouldn't have to take up more space in house than a large piece of travel luggage on a shelf.

Probably related to things like "The Nazis utilized data from routine censuses, tax returns, and municipal police registrations. In Germany, and in occupied countries like the Netherlands, this information was systematically organized. In some instances, IBM technology (via Dehomag punch card machines) was used to tabulate and sort census data to identify individuals of Jewish descent."

Ratified treaties are the supreme law of the land and he's pointing out "all lawful uses" isn't what the admin says it is.

It doesn't need to have a sleeping surface or stay sprawled out when it isn't in use. It could bunch up into a little ball and fit in the corner of your ceiling. And it doesn't need to be the size of an adult to do most household stuff, some of the unitree ones are really short in stature, a foldable step stool for reaching upper cabinets or changing lightbulbs is probably enough.

A bigger issue is whether it can really be as safe, not trip over wires, throw the baby in the trashcan, start a fire trying to make a cup of coffee and that kind of thing. Beyond accidents, lots of companies are talking about hooking these up to LLMs for planning that have horror movies in their training sets.

In some fields like comp sci, when code isn't given but the paper describes the approach, LLMs do help with the reproducibility crisis: you can ask it to reproduce the result through reimplementation by reading the paper.

If it fails you may have to double check it did properly reimplement it, but if it succeeds you do get a reproduction.

Why have they talked about this for a long time? They predicted date of code maxing out, and did so not from fitting a sigmoid or something but they predicted it would max out right during a steep part of the slope?