@cjtrowbridge: Replying to @mo.arvan AI LLM Quantization and Performance

CJ Trowbridge
CJ Trowbridge
Open In TikTok:
Region: US
Wednesday 05 June 2024 05:16:22 GMT
1449
78
13
0

Music

Download

Comments

flower_consumer
Consumer of Flowers :
This sounds very important for edge processing/tinyML? Currently alot of models are just quantized instead of training models for the hardware
2024-06-05 22:52:59
2
vegasminsoo
Minsoo Ha :
I love these optimization problems
2024-06-05 18:09:07
2
flower_consumer
Consumer of Flowers :
ohhh did Microsoft drop and cut WizardLM2? Is it any good?
2024-06-07 04:56:25
1
miketerroni
MikeF :
Yeah, but how much wood could a woodchuck woodchuck chuck if a woodchuck woodchuck could chuck wood?
2024-06-05 06:09:53
1
fbobe3
TikTok-Timelapse :
Thank you for explaining this.
2024-07-07 12:59:02
0
anisotropic.reaction
anisotropic.reaction :
Can you recommend any small but performant llms (1B params)
2024-07-04 21:14:03
0
flower_consumer
Consumer of Flowers :
If you have no method of training at a lower bit space, but say you have a 32bit-based model, could you train a 1bit model using a larger model?
2024-06-05 22:56:05
0
To see more videos from user @cjtrowbridge, please go to the Tikwm homepage.

Other Videos


About