-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
FSD v12.x (end to end AI)
- Thread starter Buckminster
- Start date
-
- Tags
- #fixphantombraking autopilot chill is not chill elon is an ass fix fsd extreme late deceleration fsd does not commit to go fsd does not maintain proper speed fsd hard and late braking fsd jerky driving fsd late deceleration fsd stops too early too quickly fsd stops too far behind line fsd tailgating in city streets
FSDtester#1
Craves Electrolytes
Agreed, if Tesla somehow decreased auto deaths to 1 per year, they would still be labeled as the death car company. Near misses and accidents avoided, don't get reported. Therefore there would be no statistics, except for that one hypothetical death.
FSDtester#1
Craves Electrolytes
No kidding, if I see one more video of someone with their hands in their lap, off the wheel, I am gonna puke. These people are begging for a collision
.@AlanSubie4Life has given countless tips on the proper way to handle nags, anticipate and react to potentially bad scenarios, how to supervise the wheel or yoke, and none of them involved his hands in his lap.
Last edited:
AlanSubie4Life
Efficiency Obsessed Member
He’s got the defeat device that uses the scroll wheels (which he claims works again in 12.3, after he complained to his buddies at Tesla).
One death per day may sound like a lot, but if ALL cars in the US were using an L5 version of Tesla FSD, then the cost of one death per day is spread over about 275 million registered cars. Assuming an average liability of $10 million, that's less than 3.6 cents per car per day. If Tesla collects 10 years of that liability that up front with every car purchase, that comes to $131.40 per car sold. Seems like a bargain.
I picked the $10 million liability number out of the air. Average liability for a traffic fatality does not seem to be publicly available. Insurance companies seem to keep that to themselves.
So far, his video is the only account that implies the scroll wheel works again to kill the nag... and I think 'implies' is the right word. Not being able to use the scroll wheel for the nag as well as insistence on using single pull for FSD are big deals to me. I'd rather stay on v11 as long as possible.
powertoold
Active Member
12.3 in a nutshell for bay area, CA:
1) less nags
2) does extremely well with "typical" city driving
3) fails with uncommonly encountered scenarios. Doesn't fail horribly, but it's annoying when it tries to get in the wrong lane and you have places to be, etc.
4) parking lot behavior is not as good as city road behavior (obviously)
5) fsdb city driving on 12.3 is *better* than fsdb highway, imagine that (once you get used to the smooth buttery city lane changes, the highway stack feels so robotic)
6) still lots of rare edge cases to train...
7) blows my mind that my 5 year old HW3 chip can run this beast
1) less nags
2) does extremely well with "typical" city driving
3) fails with uncommonly encountered scenarios. Doesn't fail horribly, but it's annoying when it tries to get in the wrong lane and you have places to be, etc.
4) parking lot behavior is not as good as city road behavior (obviously)
5) fsdb city driving on 12.3 is *better* than fsdb highway, imagine that (once you get used to the smooth buttery city lane changes, the highway stack feels so robotic)
6) still lots of rare edge cases to train...
7) blows my mind that my 5 year old HW3 chip can run this beast
14 / ~1300 PST / ~60
15 / ~2200 PST / ~60
15 / ~2330 PST / ~190
—————————————
Maybe new wave every 12-30 hours?
Hopefully this continues rolling and goes to all who are on .8 and .14. Would assume the .8 crowd would have priority since they didn’t get the recall yet but this doesn’t appear to be the case.
Last edited:
It’s diverse!
Last edited:
This is SUPER CRAZY. Since about early Beta 10.5 or so I have been bringing up the rear on all my Beta updates. I fully expected to be near last to get 12.3. What a surprise to see this. Also in ATL.
OT: Just a little Atlanta brag. In the 2020 census ATL was the 9th largest metro in the US. The 24 census estimates were release Thursday and in just 4 years we have passed Miami, Philly and D.C to become the 6th largest metro.
OT: Just a little Atlanta brag. In the 2020 census ATL was the 9th largest metro in the US. The 24 census estimates were release Thursday and in just 4 years we have passed Miami, Philly and D.C to become the 6th largest metro.
The latest wave of 12.3 rollout brings TeslaFi total pending/installed up to 272 (up from 101; roughly 2% rollout -> 5% of FSD Beta). So far completed installations are newly including Arizona, Georgia, North Carolina, Texas, Virginia; but seems to still be basically only HW3.
It’s random. I’ve been on FSDb since 10.3, across three cars (two with safety score entry into the “program”) and only once did I get an early release of new software. I was one of the first to get V11 back whenever that was.
Think you mean 10.2 since 10.3 was the infamous and scary 3 day only update that nearly derailed FSD Beta. I'm also an original day one 10.2 user (came out on a Friday and was OoT until Sunday and waited to install).
fasteddie7
Active Member
Would they need to feed video from each model vehicle that is exclusive to that model, such as if the video is from a model 3, the model x would think it was a smaller vehicle than it is and increase parking and curbing incidents?
Based on.........it will be only going to the same people who got 12.2.1. You are forever too optimistic.It MUST pass muster testing with the people who are "chosen ones" (12.2.1) and that takes some time. The BEST we can hope for is no serious bugs show up in the next couple of weeks. Then we can be hopeful of a wider rollout to some new people. If you are not carful you could end up getting it just before me which is usually near last.
This My Space post is just as applicable today as it was after the 10.2 debacle.
View attachment 1028049
Similar threads
- Article
