In my experience, cyclists travel from 15-25 in the city. Very rare to find someone going 60. Even motorcycles don’t usually go that fast. There’s a lot of people on the streets of Amsterdam.
The biggest danger here is not terribly clear. Somewhere in the intersection of people cluelessly walking into the bike lane and bicyclists hitting the clueless.
A cyclist that can do sustained 60km/h in narrow streets is an elite athlete.
Have you tried biking at 60km/h? Even on a very good bike with a gigantic gear ratio I don't think you'd manage to get close to it on flat ground.
Quick edit: just to give you some figures, if the bike + cyclist weighs in total about 80kg the cyclist needs to sustain an output of about 1000W (on an aggressive position down on the drops) without wind to ride at 60km/h on flat ground. Very few people on Earth can do that...
> cyclists doing well over 60 inside town narrow streets
I used to cycle and I don't think I could achieve 60 on flat ground like Amsterdam. That's very solid sprinting that even the best cyclists in the world can only achieve for short periods of time. On narrow Amsterdam streets with cobblestone? Impossible...
I'm fast. I've been clocked at over 40. 40 is crazy fast. Most cyclist go 15. The fast ones 20-25. The very fastest maybe 30-35. 60 is a wild exaggeration. That's what sprinters do in the Tour de France.
> And you can see here "normal" bikes clocked at 36 Km:
These bikes are not going at 36 km/h, whatever device she is using to measure it is incorrectly calibrated or simply not made to measure people on bikes.
WTF how does she measure 36km/h on the leisurely cycling people? And how does she measure the 28km/h guy when being so far from the direction he's travelling?
"Cyclist" and "someone on an illegal e-bike" are not one and the same thing.
> And you can see here "normal" bikes clocked at 36 Km:
No, those people freewheeling on a regular city bike on relatively flat ground are not doing 36 km/h. That's around 20 km/h by my somewhat well tuned eyeballs. Have you never been on a bike and measured your speed?
You're not going to be seeing cyclists going 36 km/h beyond significant downhills or sporty people on drop bar bikes putting in significant effort.
Going 60 you need to have a _really_ good bike and want to go that fast. I barely manage to keep 30 for more than 20 minutes, and I would consider myself an average cyclist on an above average bike.
You'd need a good bike and also be an elite cyclist/sprinter. I can usually manage 30kmh for an hour or so (providing there's a good tailwind and I'm going downhill), but the issue with going faster is the air resistance which is roughly proportional to the square of velocity (rolling resistance is roughly proportional to velocity so that becomes less important). Going twice as fast will mean quadrupling your power output.
Would it not make more sense, assuming the purpose is to have a quick smoke test of model quality...to do a different animal, in a different setting each time, so as to defeat any tuning for your benchmark? Then go back and do the same for other models? Keep the pelican as a side baseline?
I do that any time I'm suspicious that a model has done too well. My dream is to catch a lab that does a perfect pelican on a bicycle but is bad at other animals on other forms of transport.
Amazing that Anthropic is using this "broken and unusable" data center successfully, pushing new models and providing sufficient inference that we no longer have the absurd limits from the weeks leading up to them making that deal.
