scripts : update sync
06c86c03
ggml : build backends as libraries (llama/10256)
ce58be7e
backend cpu: add online flow for aarch64 Q4_0 GEMV/GEMM kernels (llam…
41c90650
sycl: Use syclcompat::dp4a (llama/10267)
04d1bae6
AVX BF16 and single scale quant optimizations (llama/10212)
0df66d65
cmake : restore CMakeLists.txt (llama/10256)
1d49a2e7
sync : leftovers (ggml/0)
8dffd644
ggml : fix some build issues
83c77397
ggml : remove duplicated sources from the last sync (ggml/1017)
f33c7ea0
ggml: new optimization interface (ggml/988)
adf81dc3
Make updates to fix issues with clang-cl builds while using AVX512 fl…
4b8ddfbd
ggml : optimize Q4_0 into Q4_0_X_Y repack (llama/10324)
49ca4814
vulkan: Optimize some mat-vec mul quant shaders (llama/10296)
68b198b4
llamafile : fix include path (llama/0)
7caa6b2e
ggml : fix compile warnings (llama/0)
e7263070
ggml : adapt AMX to tensor->grad removal (llama/0)
600728ea
ggml : inttypes.h -> cinttypes (llama/0)
3f1a78d6
ggml : fix possible buffer use after free in sched reserve (llama/9930)
c96434f2
CMake: default to -arch=native for CUDA build (llama/10320)
77ea626d
CUDA: remove DMMV, consolidate F16 mult mat vec (llama/10318)
8bd86888
ggml : fix undefined reference to 'getcpu' (llama/10354)
a901ba07
metal : refactor kernel args into structs (llama/10238)
dca00d83
llama : only use default buffer types for the KV cache (llama/10358)
6b4de57e
CMake: fix typo in comment [no ci] (llama/10360)
fcd8ea6a
CUDA: fix MMV kernel being used for FP16 src1 (llama/10357)
58b5fc45
metal : add `GGML_UNARY_OP_ELU` kernel (ggml/1018)
937684c8
Vulkan: Fix device info output format specifiers (llama/10366)
748d6336
metal : fox offset integer overflows in im2col (ggml/1015)
c157f624
vulkan: remove use of null initializer (llama/10372)
c4f46394
cuda : only use native when supported by cmake (llama/10389)
761d310e
sycl: Revert MUL_MAT_OP support changes (llama/10385)
8d6e30fb
vulkan: Optimize soft_max (llama/10301)
29894ef8
sycl : Add option to set the SYCL architecture for all targets (llama…
d2aaf9ec
cuda : fix CUDA_FLAGS not being applied (llama/10403)
166237d0
Add required ggml-base and backend libs to cmake pkg (llama/10407)
bfaf1fc7
ggml : sync resolve (skip) (#0)
52799f90
sync : ggml
0eddc9fc
talk-llama : sync llama.cpp
4e1f516e
whisper : adapt to new ggml (wip)
8c24c649
ggml/sched : do not skip views in pre-assignments
c8009663
whisper : use backend registry (#0)
e6114173
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Login to write a write a comment.
Login via GitHub