You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
cuda : speed-up by using CUBLAS_COMPUTE_32F instead of CUBLAS_COMPUTE_16F#3816
Closed
ggerganov wants to merge 1 commit intomasterggerganov/llama.cpp:masterfrom cuda-cublas-optsggerganov/llama.cpp:cuda-cublas-optsCopy head branch name to clipboard