How do you train a model to drive with LiDAR when the human drivers who generate the training data don’t use LiDAR?

My impression was that the state of the are was still to generate high-level data from your inputs, then react with a mixture of ML and algorithmic rules to those inputs. For example you'd use a mix of LIDAR and vision to detect that there's a pedestrian, use past frames and ML to predict the pedestrian's next position, then algorithmically check whether your vehicle's path is likely to intersect with the pedestrian's path and take appropriate action if that's the case

Under that model, LIDAR training data is easy to generate. Create situations in a lab or take recordings from real drives, label them with the high-level information contained in them and train your models to extract it. Making use of that information is the next step but doesn't fundamentally change with your sensor choice, apart from the amount of information available at different speeds, distances and driving conditions

Scan with LiDAR while manually driving.

Hell, you could even use slower offline 3d reconstruction of vision data for training, and still ultimately rely on runtime LiDAR.

But the driver isn’t reacting to any of the LiDAR readings, only what they can see, so what is the point?

You label the recordings that involve a crash, and use the lidar to avoid doing the same.

simulate is one way.

put the car in a video game and raytrace what the lidar would see

According to the paper you posted (interesting analysis), it’s not just low-vol assets, but good quality ones. Also, they would no doubt do a lot more buying of undervalued (good quality) companies if they weren’t by this point “too big to care”, so to speak.

When people say undervalued in this context, they usually mean it has a low price to earnings ratio. Intel is definitely not that - although it could still be a good bet if you have reason to believe their fortunes will greatly improve in the future. But it wouldn’t be a Buffett target in its current state.

So, best investor ever? (Not counting people who built enterprises, just people whose trade was to invest.)

I plan to buy some BRK stock. I’m sure it will be a good investment. But also, somewhat sentimentally, just to own a part of a financial masterpiece.

Not quite right. The pivot was from good companies at a great price to great companies at a good price (unless you can get a great price, but that’s unlikely).

Over the long run the latter is a better and more scalable strategy.

Think of Tesla as a well-funded pharmaceutical company that has invented a cure for a widespread ailment (call it “driving”) and now is waiting on regulatory approval.

The elegant syntax that is close to plain English sets it apart.

(It’s not so pretty these days though with all these type hints and other cruft that’s been added in the last ten years.)

Type hints are 100% optional, though.

And to be honest when you start using it, even just for simple things such as function signature, with the proper IDE it helps you catch mistakes.

I posted something similar from Grok 9 months ago, although it was “flagged” for some reason. the link still works.

https://news.ycombinator.com/item?id=43260083

the actual link on the HN page doesn't work though

Maybe it’s only visible to me https://grok.com/share/bGVnYWN5_5a484c29-ec2d-47e1-977e-c300...

Sold most of my BTC off at $120k, but kept a chunk not as an investment but as a sort of emergency fund that could be useful if for some reason I ever find myself needing to transact without using cash, bank accounts or credit cards.

They integrated it into Google search immediately so I think a lot of people will bother less with ChatGPT when a google search is just as effective.