Totally relatable pain- getting LLMs to reliably drive precise CAD operations is surprisingly hard, especially when spatial reasoning and plane/extrude/chamfer decisions go wrong 70%+ of the time.
For people looking at a different angle on the "text to 3D model" problem, I've been playing with https://www.timbr.pro lately. Not trying to replace SolidWorks precision, but great for the early fuzzy "make me something that looks roughly like X" phase before you bring it into real CAD.
I wonder whether this 30-min per day of wearing the headset, had more of meditation effect rather than actual electric interference. Would love to see placebo effect study with a mock headset.
Is it only me, or after reading this article with a lot of high-level, vague phrases and anecdotes - skipping the actual essence of many smart tricks making transformers computationally efficient - it is actually harder to grasp how transformers “really work”.
I recommend videos from Andrej Karpathy on this topic. Well delivered, clearly explaining main techniques and providing python implementation
There's also this type of articles where the first half of the article is easily understandable by a layman but then they suddenly drop a lot of jargon and math formulas and you get completely lost.
A friend once described these kin of descriptions by analogy with a recipie that went;
Recipie for buns,
First you need flour, this is a white fined grained powder that is produced from ground wheat that can be acquired by exchanging for money (a standardised convention for storing value) at a store which contains many such products. When mixed with the raising agent and other ingredients you should remove the buns from the oven when golden brown.
For this situation, if it feels worth it, I have been applying chatGPT Q&A on the jargon to bridge the gap. I haven’t read this article through yet, so can’t recommend, but in many cases it’s a super useful contextual jargon clearer.
Agreed, I have made my own shakespeare babbler following Karpathy's videos. I have a decent understanding of the structure and process but I don't really grasp how they work.
It's obvious how the error reduces, but I feel like there's something semanticly going on that isn't directly expressed in the code.
Im saving the latter half for tomorrow but so far its making sense. People have different learning styles, and I think this is lacking in the visual department. Parts like the vectors all being displayed next to the word like "cat", could have been better annotated to show where those numbers come from visually.
Gun manufacturers get some scrutiny, but it isn’t a fair comparison in that social media is engineered to keep people addicted and messes with their mental state in a destructive way. Guns are a tool abused by people with mental predisposition to do so.
The whole scene is performative nonsense anyway. Nothing will happen.
Alcohol and sugary foods are also engineered to keep people addicted and mess with their mental and physical states and are completely legal and not frowned upon.
> Alcohol and sugary foods are also engineered to keep people addicted and mess with their mental and physical states
Humans have been drinking alcohol and eating sugary foods for thousands of years. They weren’t “engineered” to do anything. They are just natural substances which humans stumbled upon and found inherently attractive (due to our biological predispositions).
Both are very energy-dense. In a premodern / early modern society in which most people engage in hard physical labour every day, that energy-density is valuable. In a late modern society in which most people are sedentary, it becomes much more harmful.
Many of the negative health consequences of both have only become known in recent decades. And several of those long-term health consequences are only apparent due to increased longevity, in centuries prior many people would have perished due to other causes long beforehand, and even deaths directly attributable to these substances would have seemed less significant against the background of generally higher mortality.
I agree—I don't think overall cookies or cheesecake are a major source of obesity, since they're really not eaten that often. You're right about soda, but in general the large amount of sugar (in whatever form) in so many foods is a problem.
The alcohol industry is thousands of years old and has gone through varying levels of regulation and deregulation. The Prohibition happened. It was a big deal.
But even then, even given their incredibly entrenched history of consumption and relatively obvious risks, they still face real civil liabilities. Hard seltzer companies, most recently, have been sued and fined for promoting their brands as healthy. A very similar case happened to the sugary drink Vitamin Water.
But, if you really want to compare the alcohol industry to social media, you’d first need to give Budweiser a way to, in real time, modify the alcohol content of the can based on the current mental state of the drinker. Do that and it might come close to be a valid comparison. But you’d still need to change the clearly labeled alcohol content on the front of the can, and instead bury it in a fourteen-page terms of service agreement. Then, also give the can the ability to simultaneously monitor you, and inform the third-party salty snack-food manufacturers that your drunk and it’s time to send you a push notification from the chips bag, as you have a 40% increased likelihood of wanting chips. That might get you there.
The prescription pain-killer and tobacco industries have paid hundreds of billions in fines for covering up their addictive properties. The list goes on.
Guns and cars don’t change people’s mental state. Some of the states with the lowest homicide rates are states like Idaho, where half of households own a gun. This is also true over time. In the 1950s in the US, the percentage of households with a gun was much higher, but homicide rates were lower. Gun ownership rates in Europe were also a lot higher in the mid 20th century, while homicide rates were not.
Does owning a gun cause suicidal thoughts? Most of Western Europe has higher suicide rates than the US: https://www.phillyvoice.com/why-suicide-rise-us-falling-most.... Eastern Europe has even higher suicide rates. And suicide rates in the US are going up as the percentage of households with a gun is going down.
It can make a difference. Someone with suicidal thoughts who has a gun, the gun can give them a quick and easy method of acting on it, with high certainty of success. Without a gun, they’d have to resort to other methods, which involve more work, have lower likelihood of success, and greater risk (and hence anxiety) of the “you don’t actually die you just end up severely disabled” outcome
For someone dead set on killing themselves, it doesn’t make a big difference. For someone who has a sudden impulsive thought of suicide, having a gun can mean they die, without one by the time they’ve worked out how to kill themselves the impulse has passed
There isn’t a simple correlation between gun ownership rates and suicide because suicide rates are determined by many factors of which gun availability is just one. But I think it is a real factor, and there is research which supports that it is, e.g. https://pubmed.ncbi.nlm.nih.gov/36652694/
No but it increases the risk of a suicide attempt and it's lethality. The link between homicide and guns is small and nebulous, the link between guns and suicide is clear and much stronger.
Have been using pandas for years and statements about polars definitely seem appealing. Especially around performance, where apply function in pandas (to iterate over rows and derive a new column) can easily be 100x slower than vanilla python. Similar with some pandas APIs culprits like making sure to reset the index after joins and other transformations
Hardware is impressive as expected - engineering porno.
One use-case that i can imagine having this headset is as a replacement for external displays. But that would need to be conditioned on great pass through and ability to use physical keyboard and mouse.
As a non-native English speaker, I often have troubles with clearly transforming abstract ideas from my head into a text representation adding grammar on top. ChatGPT can definitely help with that and increase supply of clever ideas that can be now wrapped into well written text.
I imagine you can put abstract ideas into words in your primary language more easily, so I wonder if you will then get the best English language version via an LLM or from a translation of your original text. Just idly musing since there's several variables and judging the outcome is partly subjective anyway.
For people looking at a different angle on the "text to 3D model" problem, I've been playing with https://www.timbr.pro lately. Not trying to replace SolidWorks precision, but great for the early fuzzy "make me something that looks roughly like X" phase before you bring it into real CAD.