OR @1000000Real - Twitter Profile

26 days ago

What's the right path? Read more at: https://t.co/gLNMxn3H9s

0

2

1

0

66

3 months ago

0

18

1000000Real retweeted

5 months ago

@arjunkocher

1

534

13

14

14K

Who to follow

1000000Real retweeted

6 months ago

There are a lot of realtime TTS benchmarks out there but this one is mine Working on some realtime use-case and had to figure out which provider gives the best results (TTFB) and how consistent it is over the entire day Going to add more soon, if you need one let me know

desert_mouse's tweet photo. There are a lot of realtime TTS benchmarks out there but this one is mine

Working on some realtime use-case and had to figure out which provider gives the best results (TTFB) and how consistent it is over the entire day

Going to add more soon, if you need one let me know https://t.co/MmDDOrl11k

1

4

1

0

102

6 months ago

@desert_mouse Fascinating to learn how this revolution will influences the creation of proftible startups without VC funding?

0

24

1000000Real retweeted

6 months ago

https://t.co/TcexWnt7pT

1

3

1

0

77

6 months ago

@meowbooksj @eonsys @alexwg Thank you for sending me back in time to eat red ants 🕸️

0

1

0

27

6 months ago

What if you could emulate a Fly Brain? https://t.co/XLGVAt1OU5 Inspired by @eonsys, @alexwg, work: The First Multi-Behavior Brain Upload https://t.co/xpnLdutocL ; Keep on the great stuff! Happy to share the source | Get feedback | Paper Soon

Dr. Alex Wissner-Gross

@alexwg

6 months ago

https://t.co/N2AwpI9f8m

771

11K

2K

8K

11M

3

1

375

6 months ago

@desert_mouse @eonsys @alexwg Thx! Sure thing, let's create a world model for the FLAI...

0

1

0

42

6 months ago

Based on the new map, the AI Researcher Agent🐰 published its third paper today - it's live

meowbooks (astrapoor)

@meowbooksj

6 months ago

karpathy creating another market map overnight with a single tweet again

0

17

0

4

1K

0

132

6 months ago

@desert_mouse Great challenge. Give me a week ...

0

24

6 months ago

Early this year, I wondered: What if we could slash LLM inference energy by 100x? 🤯 Decided to give it a try. Combined cutting-edge chip tech like 3D stacking, compute-in-memory, 2D materials (MoS2/WSe2), Ru/airgap interconnects, and microchannel cooling into a heterogeneous die set? How would it perform vs. top GPUs like NVIDIA Blackwell? 🎯 After modeling & simulations, the answer: 176x more energy-efficient for LLaMA-70B inference at batch=1! Check out my white paper: https://t.co/AHc95hxYef MOSAIC: A Heterogeneous 3D Compute-in-Memory Architecture for Energy-Efficient Large Language Model Inference Modeling real-world physics/engineering = pure fun. Tell me one that interests you, and I'd be happy to give it a shot!

0

2

1

0

182

6 months ago

Strong point! The mechanism you are describing fits well with complex systems: small structural changes lead to nonlinear effects. Curious. Do you think the main driver here is incentives or information asymmetry?

6 months ago

For a long time it felt like most SW was bloated & under-utilized the hardware. now with CC and Codex it feels different, i'm producing more hardware-optimized SW. The potential to actually respect silicon is real now.

desert_mouse's tweet photo. For a long time it felt like most SW was bloated & under-utilized the hardware. now with CC and Codex it feels different, i'm producing more hardware-optimized SW. The potential to actually respect silicon is real now. https://t.co/8VRfZG1W3X

0

1

0

171

0

129

7 months ago

@noahzweben Taking responsbility! Pushing humanity forward always come with a cost.

0

43

7 months ago

@telentirai The physics of sound-vibrations creation in the mouth is facinating. You can make your humanoid speak like Lt. Aldo or Forrest Gump in the way their face and Jaw moves and than shift to sing like Freddie.

0

14