10-13-2026, 06:07 AM
What worked for me was pretty simple. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at continuous batching.
trying not to rebuild again
|
NVFP4 / INT4 tradeoffs with PyTorch
|
|
« Next Oldest | Next Newest »
|
| Messages In This Thread |
|
NVFP4 / INT4 tradeoffs with PyTorch - by pixelspark - 10-12-2026, 05:31 AM
Re: NVFP4 / INT4 tradeoffs with PyTorch - by edgekite - 10-13-2026, 06:07 AM
|
Users browsing this thread: 1 Guest(s)