ggml-org/llama.cpp

Pull Requests Commits
convert : fix Qwen3-Embedding pre-tokenizer hash (#15030)
iamlemec committed 41 days ago
Verified 711d5e6f
chat : fix multiple tool_calls on hermes-2-pro (#14962)
jhen0409 committed 41 days ago
Verified f738989d
vulkan: coopmat2 mul_mat optimizations (#14934)
jeffbolznv committed 41 days ago
Verified 4cb208c9
llama-bench: rename DB table name from test to llama_bench (#15003)
yeahdongcn committed 41 days ago
Verified 3025b621
vulkan: Support ne[3]>1 in noncontig matrix-vector multiply (#15015)
jeffbolznv committed 41 days ago
Verified ec0b1880
model : support Qwen3-Embedding (#15023)
iamlemec committed 41 days ago
Verified 339bd026
server: enable token array inputs for OAI API (#15001)
JohannesGaessler committed 41 days ago
Verified f9062755
vulkan: optimizations for direct convolution (#14933)
jeffbolznv committed 41 days ago
Verified a9f7541e
CUDA: fix MMQ nwarps for AMD with warp_size==32 (#15014)
JohannesGaessler committed 42 days ago
Verified 9c35706b
vendor : update vendored copy of google/minja (#15011)
l-austenfeld committed 42 days ago
Verified c76b420e
model : add hunyuan dense (#14878)
stevenkuang-tencent committed 42 days ago
Verified 0f5ccd6f
opencl: add f16 for `add`, `sub`, `mul`, `div` (#14984)
lhez committed 42 days ago
Verified 1c872f71
ggml : Q2k interleaving implementation - x86/x64 SIMD (#14373)
Srihari-mcw committed 42 days ago
Verified baad9488
graph : fix equal_seq() check (#14986)
ggerganov committed 42 days ago
Verified ba42794c
docker : add cann build pipline (#14591)
diannaojiang committed 42 days ago
Verified 2860d479
compare-commits.sh: support both llama-bench and test-backend-ops (#14392)
yeahdongcn committed 42 days ago
Verified 484b2091
quantize : skip tensor override when in fallback mode (#14995)
EAddario committed 43 days ago
Verified daf2dd78
llama : add simple option to enable CPU for MoE weights (--cpu-moe) (#14992)
slaren committed 43 days ago
Verified a06ed5fe
Fix params bug in diffusion example (#14993)
am17an committed 43 days ago
Verified 78452405
llama : allow other bufts when overriding to CPU, add --no-repack option (#14990)
slaren committed 43 days ago
Verified d6818d06
Vulkan: Fix minor debug mode issues (#14899)
0cc4m committed 43 days ago
Verified e08a9882
mtmd : support MiniCPM-V 4.0 (#14983)
tc-mb committed 43 days ago
Verified 952a47f4
MODEL_TENSOR.SSM_DT_NORM has defined twice (#14991)
csabakecskemeti committed 43 days ago
Verified 36e5fe7b
server : implement universal assisted decoding (#12635)
g2mt committed 43 days ago
Verified 94933c8c
llama : merge build_moe_ffn_from_probs function into build_moe_ffn (#14968)
wdl339 committed 43 days ago
Verified c1dacaa9
server : add openai-style logit_bias support (#14946)
lukasstraub2 committed 43 days ago
Verified a9f77a8b
Add LLaDA 8b Diffusion model (#14771)
am17an committed 43 days ago
Verified 8a4a8562
CANN: Improve loading efficiency after converting weights to NZ format. (#14985)
hipudding committed 43 days ago
Verified 11490b36
graph : reduce splits for recurrent and hybrid models (#14825)
compilade committed 43 days ago
Verified 66625a59
opencl: add `mul_mat_f32_f32_l4_lm` and `mul_mat_f16_f32_l4_lm` (#14809)
lhez committed 43 days ago
Verified 6e672545
Terms Of Use Privacy Policy Refund Policy Impressum