| .. |
|
ggml-amx
|
60ce97c9d8
add amx kernel for gemm (#8998)
|
hai 1 ano |
|
ggml-cann
|
904837e0cb
cann: fix crash when llama-bench is running on multiple cann devices (#9627)
|
hai 1 ano |
|
ggml-cuda
|
5b359bb1e3
ggml: fix zero division in ‘dne’ calculation in CUDA COUNT_EQUAL operator when ‘ne’ is small (#10213)
|
hai 1 ano |
|
ggml-sycl
|
3bcd40b3c5
Optimize RWKV6 Operator Naming and Implement Multi-core CPU/ SYCL Acceleration (#10133)
|
hai 1 ano |
|
kompute @ 4565194ed7
|
f3f65429c4
llama : reorganize source code + improve CMake (#8006)
|
hai 1 ano |
|
kompute-shaders
|
1329c0a75e
kompute: add mul_mat_q4_k shader (#10097)
|
hai 1 ano |
|
llamafile
|
e89213492d
ggml : optimize llamafile cpu matrix multiplication for ppc64le (#10156)
|
hai 1 ano |
|
vulkan-shaders
|
8f275a7c45
ggml: Add POOL2D OP for GPU acceleration to the Vulkan backend in the MobileVLM model. (#9763)
|
hai 1 ano |
|
CMakeLists.txt
|
e89213492d
ggml : optimize llamafile cpu matrix multiplication for ppc64le (#10156)
|
hai 1 ano |
|
ggml-aarch64.c
|
9f40989351
ggml : move CPU backend to a separate file (#10144)
|
hai 1 ano |
|
ggml-aarch64.h
|
370b1f7e7a
ggml : minor naming changes (#8433)
|
hai 1 ano |
|
ggml-alloc.c
|
cd60b88bf7
ggml-alloc : remove buffer_id from leaf_alloc (ggml/987)
|
hai 1 ano |
|
ggml-amx.cpp
|
c5b0f4b5d9
llama : refactor model loader with backend registry (#10026)
|
hai 1 ano |
|
ggml-backend-impl.h
|
c5b0f4b5d9
llama : refactor model loader with backend registry (#10026)
|
hai 1 ano |
|
ggml-backend.cpp
|
9f40989351
ggml : move CPU backend to a separate file (#10144)
|
hai 1 ano |
|
ggml-blas.cpp
|
c5b0f4b5d9
llama : refactor model loader with backend registry (#10026)
|
hai 1 ano |
|
ggml-cann.cpp
|
329ed914c9
CANN: adjust backend registry refactor. (#10158)
|
hai 1 ano |
|
ggml-common.h
|
9bc6db28d0
ggml-quants : ternary packing for TriLMs and BitNet b1.58 (#8151)
|
hai 1 ano |
|
ggml-cpu-impl.h
|
23e0d70bac
ggml : move common CPU backend impl to new header (#9509)
|
hai 1 ano |
|
ggml-cpu.c
|
2319126a70
fix q4_0_8_8 format for corrupted tokens issue (#10198)
|
hai 1 ano |
|
ggml-cuda.cu
|
841f27abdb
metal : optimize FA kernels (#10171)
|
hai 1 ano |
|
ggml-impl.h
|
9f40989351
ggml : move CPU backend to a separate file (#10144)
|
hai 1 ano |
|
ggml-kompute.cpp
|
1329c0a75e
kompute: add mul_mat_q4_k shader (#10097)
|
hai 1 ano |
|
ggml-metal.m
|
39a334a9aa
metal : fix build and some more comments (#10229)
|
hai 1 ano |
|
ggml-metal.metal
|
6423c65aa8
metal : reorder write loop in mul mat kernel + style (#10231)
|
hai 1 ano |
|
ggml-quants.c
|
3407364776
Q6_K AVX improvements (#10118)
|
hai 1 ano |
|
ggml-quants.h
|
6a0f779484
ggml : add run-time detection of neon, i8mm and sve (#9331)
|
hai 1 ano |
|
ggml-rpc.cpp
|
9f40989351
ggml : move CPU backend to a separate file (#10144)
|
hai 1 ano |
|
ggml-sycl.cpp
|
3bcd40b3c5
Optimize RWKV6 Operator Naming and Implement Multi-core CPU/ SYCL Acceleration (#10133)
|
hai 1 ano |
|
ggml-vulkan.cpp
|
160687b3ed
vulkan: Fix newly added tests for permuted mul_mat and 1D im2col (#10226)
|
hai 1 ano |
|
ggml.c
|
841f27abdb
metal : optimize FA kernels (#10171)
|
hai 1 ano |
Jeff Bolz
160687b3ed
vulkan: Fix newly added tests for permuted mul_mat and 1D im2col (#10226)