Clearly you have never had to train a NN
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
OFFICIAL BUTTON WATCH
- Thread starter Drrobmiller
- Start date
diplomat33
Average guy who loves autonomous vehicles
It's a limited deployment but still a deployment IMO since some in the public are able to ride in it now.
rxlawdude
Active Member
And a couple of weeks ago you were implying that "San Franciscans who want to" can use Waymo "today" but then had to resort to tortured logic in a word salad defense of what was clearly an incorrect statement.
Which now you are backtracking on. Progress!!
diplomat33
Average guy who loves autonomous vehicles
And a couple of weeks ago you were implying that "San Franciscans who want to" can use Waymo "today" but then had to resort to tortured logic in a word salad defense of what was clearly an incorrect statement.
Which now you are backtracking on. Progress!!
I never said that San Franciscans who want to can use Waymo today. Stop twisting my words.
rxlawdude
Active Member
I'll let the reader find your posts and make that decision. You get yourself into twister positions trying to pump Waymo and denigrate FSD. Consistently. Even "disagreeing" with anyone who dares to point out the farce that is Waymo's operating plan.
diplomat33
Average guy who loves autonomous vehicles
Let's get back on topic, shall we? We are not here to debate my posts on Waymo. We are here to talk about the FSD Button.
rxlawdude
Active Member
Fair enough. There are plenty of posts in that thread that show your intense bias for, and defense of, Waymo.
That's the last time I'll type "Waymo" in this thread.
How exactly is he wrong? He is absolutely right for other reasons and that reason being that after about 6 years of AP development which is now going on 7 years. They still are no where close to a safe L5 system. So what happened to the billions of miles of data advantage? Why hasn't it shown up? According to ArkInvest which is regarded as gods in Tesla circles. They say that Tesla has over 10 billion miles of data compared to Waymo's 10 million. Yet where is the L4 let alone L5?
Last edited:
All that supposed data and yet nothing to show for it. absolutely nothing. Logic will tell you that the so-called "data advantage" is actually worthless.
You don't need a million cars. Just like you don't need 10 million cars. 10 million cars won't solve autonomous driving any quicker than 1 million cars.
diplomat33
Average guy who loves autonomous vehicles
I don't know if it is "worthless". That's a bit strong. The data does help Tesla train their NN. But certainly, the real world data is not the decisive advantage that Tesla fans believe it to be. That's because, correct me if I am wrong, but it seems to me that quality is more important than quantity after a certain point. Just having a lot of data does not necessarily translate into more progress. There is no 1:1 correlation between number of miles and rate of FSD progress.
Also, don't know if you saw this but Elon admitted that they overfit to the Bay Area:
So Tesla might have "billions of miles" of data but most of the data is confined to areas with high density of Teslas, so it is really just working great in some areas where they overfit. Other areas, where Tesla has fewer cars and less data, FSD does worse.
Also, it is ironic because I seem to recall Tesla fans saying that Waymo can't solve FSD because they are overfitting to geofenced areas and that Tesla's big data does not have that problem. Yet, we see that it is exactly Tesla that has a problem with overfitting.
rxlawdude
Active Member
Except an FSD Tesla can operate almost anywhere, while Waymo cannot. So it's ironic you bring this up.
Daniel in SD
(supervised)
You made it 7 hours without mentioning Waymo.That's the last time I'll type "Waymo" in this thread.
EVNow
Well-Known Member
They have hundreds of employees testing. YouTube beta is mainly for PR - not real testing.
rxlawdude
Active Member
Um, look who brought it up AGAIN, dude. Just pointing out logical and factual fallacies proffered by our walking Waymo ad. ;-)
powertoold
Active Member
They probably have a good idea how beta 10 performs based on their unit tests and simulations as well. In order to develop and iterate on fsd software, you need a strong testing infrastructure and shadow modes on a wide fleet. Otherwise, you're left with having to follow all your software updates with roadside assistance cars
caligula666
Member
This is merely an opinion.
For what it’s worth, we don’t know what is required from Waymo to extend to other regions, this might just be a logistical decision. Or not.
Similarly, until I see a Tesla FSD in the streets of busy Tel Aviv, circling the Arche in Paris or in the Irish narrow countryside, I’ll wait and see.
Those use cases might not be the ones you care about personally, but they are an essential part of what defines « anywhere ».
diplomat33
Average guy who loves autonomous vehicles
Yes, but FSD Beta is only L2 while Waymo is L4.
Just pointing out logical and factual fallacies proffered by Tesla fanboys.
how so? Waymo only operates in one city (two if you are very optimistic) where as you can activate fsd beta 9.3 almost anywhere so long as you have it.
I mean, yet seems like the antithesis of opinion to me and is closer to fact.
Just pointing out logical and factual fallacies proffered by Elon’s belly sweat collectors.
you guys are so cute!
Having shitty FSD development timeline does not mean there isn't a data advantage. Retraining machine learning models is the easy part, it's pretty funny when people get in a tizzy when they hear a model has to be retrained. It's just another model.fit() command.
Just because they have a data edge doesn't mean they are fully taking advantage of it yet. It doesn't guarantee success either.
I mean even now, most of their NNet models are on perception right? Most of the code isn't using neural nets, so they are less able to leverage the data set on those pieces.
I get it, you guys only want to evaluate by the final outcome. That's fine, but understand the reasons for being behind are different. For instance, I in parallel believed having more data was a potential advantage while I also believed Tesla had no chance of making any meaningful FSD when their input was on two time snapshots in the NNs (vs the "video" stream of now). I mean, having crappy perception module is a reason enough for FSD to lag behind significantly... I know you must agree with that!
Similar threads
