ggml-org/llama.cpp

Pull Requests Commits

ci : disable op offload in some tests

ggerganov committed 97 days ago

Verified 6cdda87b

llama : update worst-case graph for unified cache

ggerganov committed 98 days ago

Verified e4838046

ggml : add GGML_SCHED_NO_REALLOC option to disable reallocations in ggml_backend_sched

slaren committed 99 days ago

0710d5f0

ggml : add missing AVX512 feature checks (#17270)

angt committed 100 days ago

Verified cb623de3

metal : support I32 -> I32 copy (#17317)

ggerganov committed 100 days ago

Verified 7aaeedc0

metal : faster argsort (#17315)

ggerganov committed 100 days ago

Verified 3347e6d9

metal : add cumsum (#17305)

ggerganov committed 100 days ago

Verified 1a139644

CANN: Use smart pointers to manage ACL objects (#17238)

hipudding committed 100 days ago

Verified 2376b775

vulkan: add LOG operation support for F32 and F16 (#17183)

zayac committed 100 days ago

Verified dbed6129

vulkan: fix MMQ quantize_y condition (#17301)

0cc4m committed 100 days ago

Verified 80deff36

ci : revert #16249 (#17303)

netrunnereve committed 100 days ago

Verified 8b1c339b

metal : remove obosolete asserts (#17295)

ggerganov committed 101 days ago

Verified 416e7c7f

server : handle context overflow during decode (#17267)

ggerganov committed 101 days ago

Verified 5b2093be

opencl: fix rms_norm_mul (#17250)

lhez committed 101 days ago

Verified 52e5d421

opencl: add kernel to handle mat mul in attention to improve encoding speed (#17181)

shaofeiqi committed 101 days ago

Verified 4db56412

sycl : unify unary kernels with a generic implementation and enable wide operator support (#17213)

shani-f committed 101 days ago

Verified 72bd7321

webui: Fix clickability around chat processing statistics UI (#17278)

allozaur committed 101 days ago

Verified 22e1ce2f

webui: add OAI-Compat Harmony tool-call streaming visualization and persistence in chat UI (#16618)

ServeurpersoCom committed 101 days ago

Verified 1411d927

convert : remove unnecessary chat template patching (#17289)

CISC committed 101 days ago

Verified 662192e1

vulkan: Fuse mul_mat_id+add_id+mul and mul_mat+add+add. (#17287)

jeffbolznv committed 101 days ago

Verified 24dc769f

vulkan: Replace 16-bit unpack8 calls to work around legacy Windows AMD driver bug (#17285)

0cc4m committed 102 days ago

Verified 4dca015b

convert : use all parts in safetensors index (#17286)

CISC committed 102 days ago

Verified 9a8860cf

convert : set expert gating func in base class (#17279)

CISC committed 102 days ago

Verified 9d3ef480

mtmd-cli: Avoid logging to stdout for model loading messages in mtmd-cli (#17277)

ankurvdev committed 102 days ago

Verified c7b7db04

vulkan: implement ABS and NEG (#17245)

giuseppe committed 102 days ago

Verified 1568d13c

vulkan: Use ggml_vk_tensor_subbuffer in mul_mat_vec(id) paths (#17244)

jeffbolznv committed 102 days ago

Verified 439342ea

vulkan: skip all-negative-inf blocks in FA (#17186)

jeffbolznv committed 102 days ago

Verified 234ae7d7

vulkan: change graph_compute to be async and enable get_tensor_async (#17158)

jeffbolznv committed 102 days ago

Verified 38eaf32a

mtmd: add mtmd_log_set (#17268)

ngxson committed 103 days ago

Verified 9b17d74a

model : add AfmoeForCausalLM support (#16477)

bartowski1182 committed 103 days ago

Verified e1fcf8b0

Older