ngxson/llama.cpp

Pull Requests Commits

common : bring back --no-warmup to server

ngxson committed 1 year ago

75499af2

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

ggerganov committed 1 year ago

Verified 0cd182eb

ggml: add `GGML_SET` Metal kernel + i32 CPU kernel (ggml/1037)

PABannier committed 1 year ago

Verified a8cbab20

ggml : add `GGML_PAD_REFLECT_1D` operation (ggml/1034)

PABannier committed 1 year ago

Verified c2082d93

py : update outdated copy-paste instructions [no ci] (#10667)

danbev committed 1 year ago

Verified d405804b

Update deprecation-warning.cpp (#10619)

aryantandon01 committed 1 year ago

Verified f112d198

server : fix speculative decoding with context shift (#10641)

ggerganov committed 1 year ago

Verified 1da7b765

ggml : add predefined list of CPU backend variants to build (#10626)

slaren committed 1 year ago

Verified 59f4db10

ggml-cpu : fix HWCAP2_I8MM value (#10646)

slaren committed 1 year ago

Verified 28035408

Fix HF repo commit to clone lora test models (#10649)

ltoniazzi committed 1 year ago

Verified 253b7fde

llama: Support MiniCPM-1B (with & w/o longrope) (#10559)

JFLFY2255 committed 1 year ago

Verified 8d0cfd55

vulkan: Implement "fast divide" (mul+shift) for unary ops like copy (#10642)

jeffbolznv committed 1 year ago

Verified 2759916d

SYCL : Move to compile time oneMKL interface backend selection for NVIDIA backend (#10584)

s-Nick committed 1 year ago

Verified 40c6d79f

fix typo of README.md (#10605)

WrRan committed 1 year ago

Verified 98036d56

Avoid using __fp16 on ARM with old nvcc (#10616)

frankier committed 1 year ago

Verified cd2f37b3

Add docs for creating a static build (#10268) (#10630)

mostlygeek committed 1 year ago

Verified da6aac91

clip : add sycl support (#10574)

piDack committed 1 year ago

Verified 01e6d9bb

vulkan: optimize and reenable split_k (#10637)

jeffbolznv committed 1 year ago

Verified cc98896d

server : (web ui) Various improvements, now use vite as bundler (#10599)

ngxson committed 1 year ago

Verified 91c36c26

scripts : remove amx sync

ggerganov committed 1 year ago

1cd3df46

ggerganov committed 1 year ago

c5054718

CUDA: remove unnecessary warp reduce in FA (ggml/1032)

mahorozte committed 1 year ago

e9e661bd

feat: add `GGML_UNARY_OP_ARGMAX` Metal kernel (ggml/1019)

PABannier committed 1 year ago

efb6ae96

metal : add `GGML_OP_CONV_TRANSPOSE_1D` kernels (ggml/1026)

PABannier committed 1 year ago

667d70d1

llama : add missing LLAMA_API for llama_chat_builtin_templates (#10636)

ngxson committed 1 year ago

Verified 3b4f2e33

readme : add option, update default value, fix formatting (#10271)

pothitos committed 1 year ago

Verified 82bca225

Older