ggml-org/llama.cpp

Pull Requests Commits
convert : handle compressed-tensors quant method (#17069)
compilade committed 39 days ago
Verified 1c07c0c6
server : handle failures to restore host cache (#17078)
ggerganov committed 39 days ago
Verified cb1adf88
benches : add folder with benchmarks (#16931)
ggerganov committed 39 days ago
Verified ef1d8269
Switch to using Ubuntu 25.10 vulkan/mesa (#16497)
ericcurtin committed 39 days ago
Verified 86fde91e
vulkan: iGPU memory reporting fix (#17110)
0cc4m committed 40 days ago
Verified 7f3e9d33
vulkan: fix mmq out of bounds reads (#17108)
0cc4m committed 40 days ago
Verified 8a3519b7
vulkan: fuse mul_mat_id + mul (#17095)
jeffbolznv committed 40 days ago
Verified 80a6cf63
metal : retain src and dst buffers during async ops (#17101)
ggerganov committed 40 days ago
Verified 0750a599
arg: add --cache-list argument to list cached models (#17073)
ngxson committed 40 days ago
Verified aa3b7a90
webui: fix keyboard shortcuts for new chat & edit chat title (#17007)
chansikpark committed 40 days ago
Verified 333f2595
vulkan: Use spec constants for conv2d s/d/p and kernel W/H (#16978)
jeffbolznv committed 40 days ago
Verified 53d7d21e
server: fix correct time_ms calculation in prompt_progress (#17093)
gSUz92nc committed 40 days ago
Verified eeee367d
Revert "CUDA: add expert reduce kernel (#16857)" (#17100)
am17an committed 40 days ago
Verified 64fe17fb
CUDA: skip fusion for repeating adds in bias (#17080)
am17an committed 41 days ago
Verified c1b18768
vulkan: Increase BK to 32; use BK/4 for non-CM mul_mm.comp (#16636)
SavicStefan committed 41 days ago
Verified b8a5cfd1
ggml: disable vxe for cross-compilation by default (#16966)
AlekseiNikiforovIBM committed 41 days ago
Verified 08416ebe
vulkan: fuse rms_norm + mul + rope (+ view + set_rows) (#16977)
jeffbolznv committed 41 days ago
Verified b4e335d8
vulkan: Fix test-thread-safety crashes (#17024)
jeffbolznv committed 41 days ago
Verified d6fe40fa
CUDA: fix MMQ stream-k fixup ne1 indices (#17089)
JohannesGaessler committed 41 days ago
Verified e14e842e
ggml webgpu: faster matrix multiplication/matrix-vector multiplication (#17031)
reeselevine committed 41 days ago
Verified 647b960b
CUDA: properly handle nb00=nb02 case for cpy (#17081)
bssrdf committed 41 days ago
Verified 299f5d78
vulkan : refactor buffer handling in vk_op_f32 (#16840)
Acly committed 41 days ago
Verified ac76d362
CUDA: fix should_use_mmvf for ne11 == 1 (#17085)
JohannesGaessler committed 41 days ago
Verified 65156105
bench : cache the llama_context state at computed depth (#16944)
ggerganov committed 41 days ago
Verified 7956bb4d
hparams : add n_embd_inp() to support extended embed (#16928)
CISC committed 41 days ago
Verified 9008027a
kv-cache : pad the cache size to 256 for performance (#17046)
ggerganov committed 41 days ago
Verified 16bcc125
Revert "ggml-cpu: detect correct cpu flags for arm64 (#16229) (#16239)" (#17084)
angt committed 41 days ago
Verified 9eb9a133
ggml-cpu: detect correct cpu flags for arm64 (#16229) (#16239)
lizhenneng committed 41 days ago
Verified 7c23f3f0
server : print the samplers chain for each request (#17070)
ggerganov committed 41 days ago
Verified 8c0d6bb4
common: move download functions to download.(cpp|h) (#17059)
ngxson committed 41 days ago
Verified 5c9a18e6
Terms Of Use Privacy Policy Refund Policy Impressum