_ea1k

16 ms·

241.
by _ea1k 2y ago
If Waymo is systematically choosing safer routes, that sounds like a good thing.
242.
by _ea1k 2y ago
Right now, they are pricing them similarly to Uber and Lyft, though they do seem to offer more frequent discounts and there is no tipping yet. The experience is better too, IMO.
243.
by _ea1k 2y ago
> Your proof of concept requires a human-level intelligence being in control of the car. How is that cheaper than LiDAR? This is the core of the problem. Lidar doesn't actually change any of that, it still needs AI that's just
244.
by _ea1k 2y ago
Sure, but that test was tailor made for lidar to "succeed" even if lidar actually couldn't see the hidden obstacle. The lidar was largely blocked by the "rain", which meant it could treat it like a wall. For a real
245.
by _ea1k 2y ago
While Mobileye has focused on using lidar+cameras with each effectively serving as a form of "backup", they have also demonstrated a full city drive using nothing but cameras. I've seen enough from Waymo to see that they coul
246.
by _ea1k 2y ago
I'd argue that bitter lesson might be the other way around. Waymo has been experimenting with more end-to-end approaches and is likely to end up with something that looks more like that than a "classical control" approach, th
247.
by _ea1k 2y ago
Tesla didn't remove LIDAR, they never had it. So far, that bet is looking pretty reasonable. It seems evident at the moment that the most formidable competitors in this space could build a solid FSD product with cameras alone, with the
248.
by _ea1k 2y ago
It looks like the model itself is here: https://huggingface.co/ds4sd/SmolDocling-256M-preview It was fine tuned from this: https://huggingface.co/HuggingFaceTB/SmolVLM-256M-Instruct There's a
249.
by _ea1k 2y ago
I really want these to be able to find and even redisplay images. "Search all the hotels within 5 miles of this address and show me detailed pictures of the rooms and restrooms" Hotels would much rather show you the outside, the l
250.
by _ea1k 2y ago
I'm guessing this is a side effect of mistakes in the reinforcement learning face. It'd be really easy to build a reward model that favors passing tests, without properly measuring the quality of those tests.
251.
by _ea1k 2y ago
Yeah, IMO this is also why they can be so terrible at UI. They don't really have a feedback loop for a lot of code related issues yet. This is changing and I really expect everything to be different 12 months from now.
252.
by _ea1k 2y ago
For experimentation? Absolutely. It can often be done overnight for smaller models and reasonably sized GPUs (24GB+). It'd become a lot less practical with huge datasets, but I'd guess that a lot of fine tuning tasks aren't r
253.
by _ea1k 2y ago
I tried that once, but it is still compiling. :) Kidding... honestly that was a pretty fun distribution to play around with ~20 years ago. The documentation was really good and it was a great way to learn how a lot of the pieces of a Linux
254.
by _ea1k 2y ago
That sounds like FB Messenger and WhatsApp. I'm guessing they do some server side workaround that wouldn't work for regular text messages or Apple-y messages.
255.
by _ea1k 2y ago
Indeed. In 50 years, loading the weights and doing math should be much easier than getting some 50 year old piece of cuda code to work. Then again, CPUs will be fast enough that you'd probably just emulate amd64 and run it as CPU-only.
256.
by _ea1k 2y ago
I use it as my primary coding assistant, when I'm able to. I haven't paid for the more advanced models from others, and it seems to be the most advanced free-to-use thinking model at the moment. Aider can't use grok 3 with th
257.
by _ea1k 2y ago
It isn't just human bias that makes this problematic. Their definition of success is that a prediction of 70% should be right about 70/100 times. Give a model a big reward for "success" under this criteria and it will ha
258.
by _ea1k 2y ago
Going long on AI stocks is fairly hard. OpenAI, Antrophic, and xAI are privately held, as are most of the robotics companies and companies. Companies like Amazon, Google, or Tesla might are sort of AI investments, but they are just subsidia
259.
by _ea1k 2y ago
Indeed. TBH, I also wasn't thinking about Boston Dynamics. Their Atlas robot looks really impressive. It is easy to imagine an almost 100% automated factory in the not-too-distant future.
260.
by _ea1k 2y ago
Agreed, and good reminder that a lot of people here have probably only learned about the dot-com crash from comments and history. I remember when Cisco had a P/E in the hundreds during that era. People have forgotten just how stratosph
261.
by _ea1k 2y ago
We can debate the definition of generate, but it doesn't seem important. A key claim was that nvidia's stock price decline is inevitable, with the only question being timing. Meanwhile all of these other use cases will drive deman
262.
by _ea1k 2y ago
I came close to adding a paragraph about that. Inevitably someone would argue that there's a difference between hype peaking and 100x future growth. Regardless of whether that distinction is useful, the author makes some fairly specifi
263.
by _ea1k 2y ago
If the average person has still not ridden a self driving car, assembled by figure 02 style robots, though a drive thru with AI ordering, then we aren't even close to seeing the real peak here. >100x growth ahead for sure.
264.
by _ea1k 2y ago
They are allowed to advertise lower numbers than the EPA, and are also allowed to use different tests. Tesla typically uses the test that is most favorible to their own range rating. Some other manufacturers go to a lot of effort to make su
265.
by _ea1k 2y ago
In related news, wltp range estimates today are even worse than epa estimates back then. And the EPA standards have changed since then, so that newer Teslas are much closer to their rated ranges than they were back then. Kind of weird to se
266.
by _ea1k 2y ago
CT has v2x. I hope it comes to the other cars eventually.
267.
by _ea1k 2y ago
They also have annual sales of ~140k globally. The Y is closer to 1.2M. Even in the US, the Y sells ~2x as many as Maverick.
268.
by _ea1k 2y ago
Can I ask deep research to give me a tl;dr version?
269.
by _ea1k 2y ago
My ad blocker was detected by their site that refused entry. I find that kind of stuff more annoying than an occasional notice that they'll let me extend a warranty.
270.
by _ea1k 2y ago
I doubt total sales have peaked. Economic growth and population growth will likely keep that going up. But EVs are eating enough market share that combustion cars are likely never climbing past their peak again.

More ›