Content
@
0 reply
0 recast
0 reaction
↑ j4ck 🥶 icebreaker.xyz ↑
@j4ck.eth
pretty awesome demo from Avi Schiffman of new wearable personal AI tech you wear this necklace that ambiently listens to u all day and then u can interact with it on its knowledge graph about u and ur life https://www.instagram.com/reel/Cx2iRkuOc5R/?igshid=MWZjMTM2ODFkZg==
7 replies
1 recast
9 reactions
pugson
@pugson
not sure i believe this thing can last on a tiny battery while listening to conversations and processing them locally. but seems interesting
1 reply
0 recast
0 reaction
July
@july
Low power TinyML is coming
1 reply
0 recast
1 reaction
Griffin
@griffindotart
I’ve some friends working on optimization solutions. Will be really easy to have models on even older iPhone/android models
1 reply
0 recast
0 reaction
July
@july
I’m less interested in older iPhone/Android models as I am low ultra low powered constrained embedded devices being able to run models https://blog.st.com/stm32n6/ - this is still relatively high powered but better than STM32H7 series, Cortex M7 at 480 Hz
1 reply
0 recast
1 reaction
Griffin
@griffindotart
Oh yea that’s pretty lit
1 reply
0 recast
0 reaction
July
@july
I think there are two emerging categories: - ultra low powered constrained devices - coming up the stack ST, TI etc building embedded chips that use their HAL libraries but can write AI, TinyML, tensorflow-lite stuff - NVIDIA Jetson (Orin, AGX, Nano etc) that run embedded Linux, more for on the edge GPU capabilities
2 replies
1 recast
1 reaction
