ggml-org/llama.cpp

Pull Requests Commits

vulkan: request round-to-even for fp16 in im2col/rope_head (#10767)

jeffbolznv committed 1 year ago

Verified b685daf3

vulkan: dynamic subgroup size for the remaining k quants (#10745)

netrunnereve committed 1 year ago

Verified dafae66c

imatrix : Add imatrix to --no-context-shift (#10766)

bartowski1182 committed 1 year ago

Verified ae4b9226

CUDA: rename macros to avoid conflicts with WinAPI (#10736)

aendk committed 1 year ago

Verified 750cb3e2

server : add flag to disable the web-ui (#10762) (#10751)

eugeniosegala committed 1 year ago

Verified a86ad841

vulkan: disable spirv-opt for coopmat shaders (#10763)

jeffbolznv committed 1 year ago

Verified a05e2afc

CUDA: fix shared memory access condition for mmv (#10740)

JohannesGaessler committed 1 year ago

Verified 26a8406b

Changes to CMakePresets.json to add ninja clang target on windows (#10668)

Srihari-mcw committed 1 year ago

Verified c37fb4cf

vulkan: fix compile warnings (#10731)

jeffbolznv committed 1 year ago

Verified 3d98b4cb

cmake : simplify msvc charsets (#10672)

iboB committed 1 year ago

Verified 1a050047

server : fix format_infill (#10724)

ngxson committed 1 year ago

Verified ce8784bd

server : bring back info of final chunk in stream mode (#10722)

ngxson committed 1 year ago

Verified e52522b8

Vulkan: fix NaN in tanh.comp with AMD proprietary driver on Windows (#10723)

stduhpf committed 1 year ago

Verified 06d70147

llama : use cmake for swift build (#10525)

slaren committed 1 year ago

Verified 43ed389a

vulkan: compile a test shader in cmake to check for coopmat2 support (#10713)

jeffbolznv committed 1 year ago

Verified ecc93d05

llama : add 128k yarn context for Qwen (#10698)

robbiemu committed 1 year ago

Verified 62e84d98

server : (refactor) no more json in server_task input (#10691)

ngxson committed 1 year ago

Verified 3573fa8e

ggml : disable iq4_nl interleave size 8 (#10709)

ggerganov committed 1 year ago

Verified d9c3ba2b

server : various fixes (#10704)

ggerganov committed 1 year ago

Verified ce4a7b84

ggml : refactor online repacking (#10446)

Djip007 committed 1 year ago

Verified 19d8762a

server : fix free of spec context and batch (#10651)

ggerganov committed 1 year ago

Verified c2a16c0b

Vulkan: VK_KHR_cooperative_matrix support to speed up prompt processing (#10597)

0cc4m committed 1 year ago

Verified 3df784b3

metal : Extend how Llama.cpp locates metal resources (#10676)

ormandi committed 1 year ago

Verified 86a19349

convert : add support for Roberta embeddings (#10695)

Ssukriti committed 1 year ago

Verified 784a14aa

convert : add custom attention mapping

ggerganov committed 1 year ago

Verified c5ede384

common : bring back --no-warmup to server (#10686)

ngxson committed 1 year ago

Verified f162d45a

server : (refactoring) do not rely on JSON internally (#10643)

ngxson committed 1 year ago

Verified 6c5bc062

fix(server) : not show alert when DONE is received (#10674)

pminev committed 1 year ago

Verified 7736837d

vulkan: Add VK_NV_cooperative_matrix2 support for mul_mat and flash attention (#10206)

jeffbolznv committed 1 year ago

Verified c9c6e01d

llama : add Minerva 7B model support (#10673)

Riccorl committed 1 year ago

Verified 6fe62478

Newer Older