ggerganov/llama.cpp

Pull Requests Commits

ci : add sanitizer runs for server

ggerganov committed 5 hours ago

Verified dd3fb4b3

metal : minor cleanup (#19251)

ggerganov committed 6 hours ago

Verified c55bce41

CUDA: Fix loop unrolling for BW in mul_mat_q_stream_k_fixup (#19053)

ORippler committed 7 hours ago

Verified 1f1e57f2

ggml: added cleanups in ggml_quantize_free (#19278)

noctrex committed 11 hours ago

Verified e9a859db

cuda : revert CUDA_SCALE_LAUNCH_QUEUES override until investigated (#19227)

gaugarg-nv committed 11 hours ago

Verified 41e3f026

vocab: add Falcon-H1-Tiny-Coder FIM tokens (#19249)

vhsw committed 11 hours ago

Verified 1efb5f7a

spec : simplify time measurement using common_time_meas (#19262)

ggerganov committed 11 hours ago

Verified aeb827a3

opencl: refactor some ops, concat, repeat, tanh and scale (#19226)

lhez committed 18 hours ago

Verified 91ea44e8

jinja : add missing 'in' test to template engine (#19004) (#19239)

sidmohan0 committed 22 hours ago

Verified 0dfcd3b6

mtmd: add min/max pixels gguf metadata (#19273)

ngxson committed 22 hours ago

Verified 07a7412a

ggml-cpu: FA split across kv for faster TG (#19209)

am17an committed 1 day ago

Verified 9f682fb6

server: print actual model name in 'model not found" error (#19117)

teto committed 1 day ago

Verified a3fa0358

ci: add test-backend-ops test for CPU (#19268)

am17an committed 1 day ago

Verified 15818ac4

Remove support for Nvidia & AMD GPU, because the oneAPI plugin for Nvidia & AMD GPU is unavailable: download/installation channels are out of work. (#19246)

arthw committed 1 day ago

Verified bf38346d

sycl: implement GGML_OP_TOP_K (#19242)

tdevelope committed 1 day ago

Verified 4d5e9726

metal : support virtual devices (#18919)

ggerganov committed 1 day ago

Verified 6fdddb49

model-conversion : add debug option to conversion script (#19265)

danbev committed 1 day ago

Verified 6156ae51

ggml-backend: fix async set/get fallback sync (#19179)

JohannesGaessler committed 1 day ago

Verified 59377a6c

authors : update (#19263)

ggerganov committed 1 day ago

Verified 1239267c

docs : Minor cleanups (#19252)

ckastner committed 1 day ago

Verified 7a4ca3cb

spec : various improvements ton ngram-map + docs (#19253)

srogmann committed 1 day ago

Verified b4d05a3d

Remove pipeline cache mutexes (#19195)

nikhilJain17 committed 1 day ago

Verified 2dc3ce21

Bump cmake max version (needed for Windows on Snapdragon builds) (#19188)

max-krasnyansky committed 1 day ago

Verified 3bc8d2cf

nix: fix allowUnfreePredicate for packages with multiple licenses (#19237)

typedrat committed 1 day ago

Verified 8a98ba45

create test.sh to enhance the parameters for testing, update the guide, rm useless script (#19243)

arthw committed 2 days ago

Verified 2634ed20

nix: fix nix develop .#python-scripts (#19218)

teto committed 3 days ago

Verified 41ea2614

ggml-hexagon: flash-attention and reduce-sum optimizations (#19141)

chraac committed 3 days ago

Verified 89f10baa

quantize: add option --tensor-type-file to llama-quantize (#18572)

EugeoSynthesisThirtyTwo committed 3 days ago

Verified 3dd95914

mtmd: support MiniCPM-o 4.5(vision only) (#19211)

tc-mb committed 3 days ago

Verified ec6c7421

lookup, lookahead: fix crash when n_ctx not specified (#18729)

pestopoppa committed 3 days ago

Verified 14883391

Older