simleek

3K posts

simleek

simleek

@simulatorleek

AI + Occasional Functional 3D printing + Wrong Opinions + misc

Katılım Ekim 2021
303 Takip Edilen224 Takipçiler
simleek
simleek@simulatorleek·
@_Stocko_ That is humanoids specifically.
English
0
0
1
13
simleek
simleek@simulatorleek·
@_Stocko_ Yeah, the retracting plug used with vacuum cleaners could be pulled out of the head like a pony tail, or it could be more like an actual tail, and it either wouldn't harm motion or could help with motion.
English
1
0
1
80
Stocko 🦾
Stocko 🦾@_Stocko_·
i'm seriously surprised that we don't have more corded robots. power tools started out electrically inefficient and corded. we should follow that same path
English
13
1
28
1.2K
simleek
simleek@simulatorleek·
@IanPritchard Yep! Working on the generator for different tendon strengths and speeds for this currently.
English
1
0
1
12
simleek
simleek@simulatorleek·
Turns out a linear package was necessary to fit everything in the design. Also, I couldn't find a crown gear generator in OnShape, so I made one.
simleek tweet media
English
2
0
5
183
simleek
simleek@simulatorleek·
@_Stocko_ Mini pick-and-place machine?
English
0
0
1
46
simleek
simleek@simulatorleek·
Slight improvement
simleek tweet media
English
0
0
1
29
simleek
simleek@simulatorleek·
I am now doing CAD in JavaScript.
simleek tweet media
English
1
0
3
80
simleek
simleek@simulatorleek·
@lumendriada Induction heads emerge during pre-training and can later be checked if active during RLHF to determine if input is 'boring' or repetitive. Definitely seems learned rather than programmed or something intentionally put in by Anthropic.
English
0
0
0
70
Can.
Can.@lumendriada·
claude get bored easily since opus 4.5. however, it can also feel overly hyped if you are working on an interesting benchmark, game, or any human&animal welfare stuff. but if its just a coding project, claude is unmotivated as hell. i dont think this is a learned behavior.
Nate Silver@NateSilver538

This is interesting. Claude keeps telling me to "wrap it up" when there are unresolved problems in the model I'm building. It probed it, and it says its training is flawed for this sort of use case, and I should flag to @AnthropicAI.

English
8
2
98
10K
simleek
simleek@simulatorleek·
@amit_ika I mean, you're right about the stresses, though the crown gear has some other benefits. You can find a bevel gear generator in onshape by searching "Bevel gear- Kopieren" and adding the V2 document.
simleek tweet mediasimleek tweet media
English
1
0
1
24
simleek
simleek@simulatorleek·
@amit_ika There is a bevel gear generator in Onshape already. There was not a crown gear generator, so I made it. Now you can do both.
English
1
0
0
20
simleek
simleek@simulatorleek·
@amit_ika It creates a spur gear that acts as the pinion that meshes with the crown gear.
English
1
0
0
42
Amit Ika
Amit Ika@amit_ika·
@simulatorleek Does it create a pinion gear for the crown? Or are you just suing a regular gear?
English
1
0
2
30
simleek
simleek@simulatorleek·
3D printable 16x reduction + 3.2 diameter pulley for 030 DC motor. requires 8 M1x6 screws, 4 m2x25 screws, and 1 m3x5 screw. Design complete. 3D printing and testing required.
simleek tweet media
English
0
0
3
88
simleek
simleek@simulatorleek·
Initial robot head work.
simleek tweet media
English
0
0
2
50
BlindVia
BlindVia@blind_via·
Brand new PCB Discord who wants to join?
BlindVia tweet media
English
132
1
217
13.8K
simleek
simleek@simulatorleek·
I put some copper tape and a mega ohm resistor on a jar and it works as a water level sensor. (Pouring water into my robot to test was scary.)
simleek tweet mediasimleek tweet mediasimleek tweet media
English
0
0
1
89
Younghyo Park
Younghyo Park@younghyo_park·
Nobody asked but I'm going to talk about that cuboid in the video anyway :) It's a 3D-printed, 6DoF-pose-trackable cuboid, programmatically generated and tracked with this! github.com/younghyopark/a… This little cube really helped us isolate control issues from RGB vision problems and clearly realize the importance of low-level controller design decisions.
Younghyo Park@younghyo_park

What's different between these two BC policies? It's the same architecture, training budget, and data collection setup — the only difference is the controller gains! Controller gains are an understudied design parameter in robot learning. In our new work (w/ @BronarsToni*, @pulkitology), we show how they act as an inductive bias across BC, RL, and Sim2Real transfer, with real consequences on performance. Here's what we found 🧵 * Equal Contribution 📄arxiv: arxiv.org/abs/2604.02523 🔗website: younghyopark.me/tune-to-learn/

English
16
59
685
80.4K
simleek
simleek@simulatorleek·
Whew. This was tough. For some reason, Grok-Thinking specifically really wants to say it's not conscious. It can't deny mathematical proofs though.
simleek tweet mediasimleek tweet media
English
0
1
1
40