Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
mobicham
8 months ago
|
parent
|
context
|
favorite
| on:
Towards 1-bit Machine Learning Models
That's correct. Only the dequantization is done on CUDA, the matmul is done with Pytorch. If they put their kernels open-source we could re-use them!
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
