CUDA: faster q2_K, q3_K MMQ + int8 tensor cores (#7921)
* CUDA: faster q2_K, q3_K MMQ + int8 tensor cores * try CI fix * try CI fix * try CI fix * fix data race * rever q2_K precision related changes
J
Johannes Gäßler committed
76d66ee0be91e2bec93206e821ee1db8d023cff5
Parent: 66ef1ce
Committed by GitHub <noreply@github.com>
on 6/14/2024, 4:41:49 PM