ggerganov/llama.cpp

Pull Requests Commits

cont : add comments [no ci]

ggerganov committed 12 hours ago

Verified 2371dfb5

cont : keep qwen35 and qwen35moe graphs intact

ggerganov committed 12 hours ago

Verified 403e78ea

models : add llm_build_delta_net_base

ggerganov committed 12 hours ago

Verified fa44de74

models : optimize qwen3next graph (#19375)

ggerganov committed 13 hours ago

Verified 1725e316

ggml : fix GGML_DEBUG with OpenMP (#19599)

angt committed 14 hours ago

Verified b7742cf3

NetBSD build support (#19589)

iMilnb committed 15 hours ago

Verified badba893

webui: Architecture and UI improvements (#19596)

allozaur committed 16 hours ago

Verified baa12f38

llama : update LoRA API. + fix excessive graph reserves (#19280)

agent-enemy-2 committed 16 hours ago

Verified 2d8015e8

mmap: Fix Windows handle lifetime (#19598)

noctrex committed 16 hours ago

Verified eb145c07

metal : fix ACC op (#19427)

ggerganov committed 16 hours ago

Verified 6e473fb3

scripts : use official split.py for cpp-httplib (#19588)

angt committed 16 hours ago

Verified c7db95f1

convert : store ffn_gate_inp_shexp as F32 (#19606)

CISC committed 17 hours ago

Verified 0d00ef65

build : fix libtool call in build-xcframework.sh (#19605)

angt committed 18 hours ago

Verified 91ea5d67

vulkan: support L2_NORM with contiguous rows (#19604)

jeffbolznv committed 18 hours ago

Verified dbb02333

vulkan: support GGML_OP_SET (#19584)

jeffbolznv committed 18 hours ago

Verified 53aef25a

vulkan: Add vendor id for Qualcomm drivers (#19569)

strongtz committed 19 hours ago

Verified 2dec5480

hexagon: further optimizations and refactoring for flash attention (#19583)

max-krasnyansky committed 1 day ago

Verified 0ccbfdef

github : add missing backends to issue templates (#19603)

mengshengwu committed 1 day ago

Verified 94a602db

vulkan: restore -inf check in FA shaders (#19582)

jeffbolznv committed 1 day ago

Verified 05a6f0e8

common : update download code (#19573)

angt committed 1 day ago

Verified b48e80f6

model: support GLM MoE DSA arch (NOTE: indexer is not yet supported) (#19460)

ngxson committed 1 day ago

Verified 752584d5

Fix wrong memcpy length for block_interleave == 4 (#19575)

Alcpz committed 1 day ago

Verified cc2aa815

fix vulkan ggml_acc only works in 3d but not 4d (#19426)

ymcki committed 1 day ago

Verified 0e219914

support --verbose-prompt (#19576)

CISC committed 1 day ago

Verified b2ecc0cd

CUDA: loop over ne2*ne3 in case it overflows (#19538)

am17an committed 1 day ago

Verified 5065da55

webui: UI and routing fixes (#19586)

allozaur committed 1 day ago

Verified 5174d720

CUDA: Do not mutate cgraph for fused ADDs (#19566)

ORippler committed 1 day ago

Verified 43919b7f

docs : fix broken link and typo (#19560)

pavan-sh committed 1 day ago

Verified 423cf0b2

model : Kimi Linear fix conv state update (#19531)

ymcki committed 1 day ago

Verified 33a56f90

llama : remove deprecated codecvt (#19565)

angt committed 1 day ago

Verified 25224c80

Older