ggerganov/llama.cpp

Pull Requests Commits

server : fix infill prompt format

ggerganov committed 170 days ago

Verified b8d1b1a5

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

ngxson committed 170 days ago

Verified e52522b8

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

stduhpf committed 170 days ago

Verified 06d70147

llama : use cmake for swift build (#10525)

slaren committed 170 days ago

Verified 43ed389a

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

jeffbolznv committed 170 days ago

Verified ecc93d05

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

robbiemu committed 171 days ago

Verified 62e84d98

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

ngxson committed 171 days ago

Verified 3573fa8e

ggml : disable iq4_nl interleave size 8 (#10709)

ggerganov committed 171 days ago

Verified d9c3ba2b

server : various fixes (#10704)

ggerganov committed 171 days ago

Verified ce4a7b84

ggml : refactor online repacking (#10446)

Djip007 committed 171 days ago

Verified 19d8762a

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

ggerganov committed 171 days ago

Verified c2a16c0b

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

0cc4m committed 171 days ago

Verified 3df784b3

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

ormandi committed 171 days ago

Verified 86a19349

convert : add support for Roberta embeddings (#10695)

Ssukriti committed 171 days ago

Verified 784a14aa

convert : add custom attention mapping

ggerganov committed 172 days ago

Verified c5ede384

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

ngxson committed 172 days ago

Verified f162d45a

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

ngxson committed 172 days ago

Verified 6c5bc062

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

pminev committed 173 days ago

Verified 7736837d

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

jeffbolznv committed 173 days ago

Verified c9c6e01d

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

Riccorl committed 173 days ago

Verified 6fe62478

ggerganov committed 173 days ago

Verified 0cd182eb

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

PABannier committed 173 days ago

Verified a8cbab20

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

PABannier committed 173 days ago

Verified c2082d93

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

danbev committed 173 days ago

Verified d405804b

Update deprecation-warning.cpp (#10619)

aryantandon01 committed 174 days ago

Verified f112d198

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

ggerganov committed 174 days ago

Verified 1da7b765

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

slaren committed 174 days ago

Verified 59f4db10

ggml-cpu : fix HWCAP2_I8MM value (#10646)

slaren committed 174 days ago

Verified 28035408

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

ltoniazzi committed 174 days ago

Verified 253b7fde

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

JFLFY2255 committed 174 days ago

Verified 8d0cfd55

Older