| .. |
|
ggml-blas
|
4d3d455d3c
sync : whisper.cpp (ggml/1359)
|
4 months ago |
|
ggml-cann
|
3479efd112
CANN: Improve device ID handling and aclnnArange checks (#16752)
|
3 months ago |
|
ggml-cpu
|
d38d9f0877
ggml: add s390x cpu-feats (#16774)
|
2 months ago |
|
ggml-cuda
|
7db35a7958
CUDA: add FLOOR, CEIL, ROUND, TRUNC unary ops (#16917)
|
2 months ago |
|
ggml-hexagon
|
13002a0896
ggml-hexagon: respect input size when getting/setting tensor data (#16836)
|
2 months ago |
|
ggml-hip
|
80d28f104c
HIP: fix AMDGPU_TARGETS, update documentation (#16803)
|
3 months ago |
|
ggml-metal
|
2f966b8ed8
clip : use FA (#16837)
|
2 months ago |
|
ggml-musa
|
11f0af5504
CUDA: faster tile FA, add oob checks, more HSs (#16492)
|
3 months ago |
|
ggml-opencl
|
9984cbb61d
opencl: fix boundary handling for mul_mm (#16875)
|
2 months ago |
|
ggml-rpc
|
41386cf365
rpc : report actual free memory (#16616)
|
3 months ago |
|
ggml-sycl
|
7e994168b1
SYCL: optimized repeat_back kernel (3× fewer asm instructions, 2× faster)Feature/sycl repeat back opt (#16869)
|
2 months ago |
|
ggml-vulkan
|
5d8bb900bc
vulkan: Fix multi_add invalid descriptor usage (#16899)
|
2 months ago |
|
ggml-webgpu
|
d261223d24
model: add support for qwen3vl series (#16780)
|
2 months ago |
|
ggml-zdnn
|
264f1b5187
zdnn: refactor codebase + add docs (#16178)
|
4 months ago |
|
CMakeLists.txt
|
d38d9f0877
ggml: add s390x cpu-feats (#16774)
|
2 months ago |
|
ggml-alloc.c
|
3470a5c891
ggml-alloc : make gallocr prefer chunks that allow memory reuse (#16788)
|
3 months ago |
|
ggml-backend-impl.h
|
898acba681
rpc : add support for multiple devices (#16276)
|
3 months ago |
|
ggml-backend-reg.cpp
|
63d2fc46e1
Add experimental ggml-hexagon backend for the Hexagon NPU (#16547)
|
3 months ago |
|
ggml-backend.cpp
|
e789095502
llama: print memory breakdown on exit (#15860)
|
4 months ago |
|
ggml-common.h
|
fd1234cb46
llama : add gpt-oss (#15091)
|
5 months ago |
|
ggml-impl.h
|
10fcc41290
vulkan: Update topk_moe fusion to handle gpt's late softmax (#16656)
|
3 months ago |
|
ggml-opt.cpp
|
5cdb27e091
finetune: SGD optimizer, more CLI args (#13873)
|
5 months ago |
|
ggml-quants.c
|
f6b4af3d04
ggml : fix uninitialized is_on_grid in quantize_row_iq3_xxs_impl (#15928)
|
4 months ago |
|
ggml-quants.h
|
fd1234cb46
llama : add gpt-oss (#15091)
|
5 months ago |
|
ggml-threading.cpp
|
ae8de6d50a
ggml : build backends as libraries (#10256)
|
1 year ago |
|
ggml-threading.h
|
cb13ef85a4
remove CMAKE_WINDOWS_EXPORT_ALL_SYMBOLS (#10797)
|
1 year ago |
|
ggml.c
|
4926419c4d
ggml: add ggml_can_fuse_subgraph (#16662)
|
3 months ago |
|
ggml.cpp
|
fedf034a98
ggml : Print backtrace on uncaught C++ exceptions (ggml/1232)
|
8 months ago |
|
gguf.cpp
|
a81283820a
gguf: gguf_writer refactor (#15691)
|
4 months ago |
shani-f
7e994168b1
SYCL: optimized repeat_back kernel (3× fewer asm instructions, 2× faster)Feature/sycl repeat back opt (#16869)