 0cc4m
|
3d7ebf6312
Vulkan Mixture of Experts (MoE) support (#7628)
|
1 год назад |
 k.h.lai
|
56411a950f
vulkan: properly initialize vulkan devices for LLAMA_SPLIT_MODE_NONE (#7552)
|
1 год назад |
|
0cc4m
|
1b1e27cb49
Update vulkan rope implementation to support frequency factors (#7475)
|
1 год назад |
 liuwei-git
|
201cc11afa
llama : add phi3 128K model support (#7225)
|
1 год назад |
|
0cc4m
|
f030ec1f7a
Vulkan Embedding Fix (#7360)
|
1 год назад |
|
0cc4m
|
c1b295eea5
Update and fix Vulkan soft_max and argsort implementations (#7237)
|
1 год назад |
 Georgi Gerganov
|
9cb317f77e
ggml : full ALiBi support (#7192)
|
1 год назад |
|
0cc4m
|
befddd0f15
Vulkan Bugfixes and Improvements (#7084)
|
1 год назад |
|
Georgi Gerganov
|
9c67c2773d
ggml : add Flash Attention (#5021)
|
1 год назад |
|
0cc4m
|
ba0c7c70ab
Vulkan k-quant mmq and ggml-backend offload functionality (#6155)
|
1 год назад |
 compilade
|
557410b8f0
llama : greatly reduce output buffer memory usage (#6122)
|
1 год назад |
 slaren
|
2bf8d0f7c4
backend : offload large batches to GPU (#6083)
|
1 год назад |
 GainLee
|
dc0f612548
ggml:fix finding transfer queue family index error (#6094)
|
1 год назад |
|
slaren
|
f30ea47a87
llama : add pipeline parallelism support (#6017)
|
1 год назад |
|
Georgi Gerganov
|
5b09797321
ggml : remove old quantization functions (#5942)
|
1 год назад |
|
0cc4m
|
61d1c88e15
Vulkan Improvements (#5835)
|
1 год назад |
 Michael Podvitskiy
|
9fa2627347
ggml : introduce ggml_status (ggml/750)
|
1 год назад |
 ddpasa
|
c2224f003b
ggml-vulkan: fix VULKAN_CHECK_RESULTS flag, which was previously broken (#5813)
|
1 год назад |
 Eve
|
317709b2a8
make portability_enumeration_ext apple only (#5757)
|
1 год назад |
 UEXTM.com
|
5f70671856
Introduce backend GUIDs (ggml/743)
|
1 год назад |
|
Georgi Gerganov
|
ab336a9d5e
code : normalize enum names (#5697)
|
1 год назад |
|
0cc4m
|
bb9dcd560a
Refactor validation and enumeration platform checks into functions to clean up ggml_vk_instance_init()
|
1 год назад |
|
0cc4m
|
f50db6ae0b
Add check for VK_KHR_portability_enumeration for MoltenVK support
|
1 год назад |
 Mathijs de Bruin
|
d8c054517d
Add preprocessor checks for Apple devices.
|
1 год назад |
|
Mathijs de Bruin
|
42f664a382
Resolve ErrorIncompatibleDriver with Vulkan on MacOS.
|
1 год назад |
|
Georgi Gerganov
|
5bf2b94dd4
cmake : fix VULKAN and ROCm builds (#5525)
|
1 год назад |
 Neuman Vong
|
704359e299
vulkan: Find optimal memory type but with fallback (#5381)
|
1 год назад |
 Sergio López
|
c88c74f967
vulkan: only use M-sized matmul on Apple GPUs (#5412)
|
1 год назад |
|
0cc4m
|
44fbe34360
Fix Vulkan crash on APUs with very little device memory (#5424)
|
1 год назад |
|
0cc4m
|
ee1628bdfe
Basic Vulkan Multi-GPU implementation (#5321)
|
1 год назад |
