CUDA: faster dequantize kernels for Q4_0 and Q4_1 (#4938) #16
YannFolletbuild.yml
on: push
Matrix: windows-latest-cmake-cublas
Matrix: windows-latest-cmake
ubuntu-focal-make
1m 37s
ubuntu-latest-cmake
1m 43s
macOS-latest-make
2m 14s
macOS-latest-cmake
3m 32s
macOS-latest-cmake-ios
1m 28s
macOS-latest-cmake-tvos
1m 27s
ios-xcode-build
1m 7s
Matrix: macOS-latest-swift
Matrix: ubuntu-latest-cmake-mpi
Matrix: ubuntu-latest-cmake-sanitizer
release
19s
Annotations
2 errors
|
windows-latest-cmake (avx512, -DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DLLAMA_AVX512=ON -DBUIL...
Process completed with exit code 1.
|
|
release
Resource not accessible by integration
|
Artifacts
Produced during runtime
| Name | Size | |
|---|---|---|
|
artifact
Expired
|
751 MB |
|