ggml-org/llama.cpp

Pull Requests Commits

llama : remove quantization sanity check (#17788)

danbev committed 181 days ago

Verified 444f00b0

vulkan: Use one row per workgroup for f32 mmv (#17711)

jeffbolznv committed 181 days ago

Verified 2960eb29

convert: support Mistral 3 Large MoE (#17730)

ngxson committed 181 days ago

Verified dbc15a79

vulkan: support solve_tri with larger N/K values (#17781)

jeffbolznv committed 181 days ago

Verified c6c5e859

contrib : stale PRs (#17803)

ggerganov committed 181 days ago

Verified 8e5f4987

metal : fix build(#17799)

ggerganov committed 181 days ago

Verified 8ce774a1

vulkan: Replace deprecated VK_EXT_validation_features (#17637)

rillomas committed 181 days ago

Verified 67788f68

vulkan: Fix mismatch in TOPK_MOE unit test (#17541)

rillomas committed 181 days ago

Verified d8c0a7b0

vulkan: add more num_blocks instantiations in rms_norm (#17701)

jeffbolznv committed 181 days ago

Verified 933414c0

vulkan: fix top_k bug when there are ties in the input (#17659)

jeffbolznv committed 181 days ago

Verified a0f3897d

vulkan : support conv-2d with large output size (#17685)

Acly committed 181 days ago

Verified e15cd06a

ggml webgpu: unary op suppport, code refactoring, ops support (#17764)

reeselevine committed 181 days ago

Verified fd57b24c

vulkan: enable mmvq for q2_k on NVIDIA (#17675)

jeffbolznv committed 181 days ago

Verified 6ab0d649

vulkan: set all memory allocations to high priority (#17624)

jeffbolznv committed 181 days ago

Verified 93bb9266

rpc : fix alloc size logic (#17116)

ggerganov committed 181 days ago

Verified 8160b38a

metal : add residency sets keep-alive heartbeat (#17766)

ggerganov committed 181 days ago

Verified c41bde6f

HIP : fix RDNA4 build (#17792)

JohannesGaessler committed 182 days ago

Verified 6016d0bd

fix: prevent segfault in tokenizer on highly repetitive input (#17786)

ServeurpersoCom committed 182 days ago

Verified 1be97831

ci : fix winget workflow (#17790)

angt committed 182 days ago

Verified a6cfc212

Q4/Q8 Tiled Gemm Optimization. (#16999)

shalinib-ibm committed 182 days ago

Verified 3a0d1053

Add pwilkin to CODEOWNERS for chat files (#17789)

pwilkin committed 182 days ago

Verified 66489896

CUDA: fix FA VKQ accumulator overflow (#17746)

JohannesGaessler committed 182 days ago

Verified e95d0bc8

HIP: enable WMMA-MMQ INT kernels for RDNA 3 (#17576)

jiachengjason committed 182 days ago

Verified 668ed765

ci : transform release binary root dir in tar to llama-bXXXX (#17773)

CISC committed 182 days ago

Verified 03d9a77b

docs : update ops.md (Metal, BLAS) (#17768)

gabe-l-hart committed 182 days ago

Verified 3143a755

Add support for CUMSUM and TRI for CUDA. (#17584)

pwilkin committed 182 days ago

Verified 96fe9bad

metal: TRI, FILL, EXPM1, SOFTPLUS (#16623)

gabe-l-hart committed 183 days ago

Verified bde188d6

server: strip content-length header on proxy (#17734)

ngxson committed 183 days ago

Verified 9d022996

server: move msg diffs tracking to HTTP thread (#17740)

ngxson committed 183 days ago

Verified c4c10bfb

examples : add missing code block end marker [no ci] (#17756)

danbev committed 183 days ago

Verified 817d743c

Newer Older