vogel
Region: US
Wednesday 16 September 2026 03:04:18 GMT
16202
403
23
37
Music
Download
Comments
Sean :
I remember when that was called "machine learning."
2026-09-16 15:25:33
63
Time To Wake up :
I still don't see a use for this.
2026-09-17 09:20:02
0
KayoNai[加代ない] :
does the UI not know basic math..? 1s / 200ms = 5? it's 5 per second. 40 per second is 25 ms. That UI is lying.
2026-09-16 06:14:03
4
Geoff Cross :
this seems like just algorithmic AI...which we've already had
2026-09-16 14:14:11
12
Patrick Lewis :
FYI I believe the point is you didn’t need a huge training set to get this. You’re up and running on a new set of data without training the smart thing to do would be to use this to train a traditional ml model.
2026-09-17 01:32:40
1
Red :
Jev is now ai?
2026-09-16 12:11:10
0
Alika Wejrowski :
The future of ai is the most efficient, fast model
2026-09-16 10:14:55
0
Stove :
by classifying generic requests well, you can direct the request to the right SLM for that specific task. you can then get high quality output with a fraction of the cost and time of a frontier model. basically this is what MoE models do, but splitting into fully separate models would mean being able to mix providers. intelligence is truly commoditized at that point
2026-09-16 19:22:26
5
Jimmyhaxx :
thats just ~11 emails per agent. its not alot to be impressed tbh
2026-09-16 17:06:33
1
Robert :
Okay now review all 1,500 email classifications, ensure every category is correct, and run it again and tell us the precision, recall, F1
2026-09-16 20:42:05
0
cheese lord :
Not a cofounder, purposely using misleading language to farm hype. Co-inventor, aka “an engineer on the initial ChatGPT model (3.5)”
2026-09-17 05:24:16
0
To see more videos from user @vogeldev, please go to the Tikwm
homepage.
