ggerganov/llama.cpp

Pull Requests Commits

Merge branch 'master' into compilade/imatrix-neutral-prior

compilade committed 33 days ago

95ae9982

Revert "quantize : assume the neutral prior is equal imatrix weights"

compilade committed 33 days ago

f45b59a5

vendor : update cpp-httplib to 0.37.1 (#20390)

cabelo committed 33 days ago

Verified 557fe2d9

tests : use `reasoning` instead of `reasoning_budget` in server tests (#20432)

pwilkin committed 33 days ago

Verified 0e810413

test-backend-ops: allow loading tests from file and parsing model operators into file (#19896)

0cc4m committed 33 days ago

Verified 128142fe

common : update completion executables list [no ci] (#19934)

danbev committed 33 days ago

Verified 6de1bc63

grammar: Fix grammar root symbol check (#19761)

AsbjornOlling committed 33 days ago

Verified 0a10c34d

vulkan: add GATED_DELTA_NET op support (#20334)

ProgenyAlpha committed 33 days ago

Verified deee2386

convert : better mtp check and fix return [no ci] (#20419)

CISC committed 34 days ago

Verified c3e3f9e5

vulkan: fix SSM_CONV PP scaling with large ubatch sizes (#20379)

ProgenyAlpha committed 34 days ago

Verified 40c550d4

New conversations now auto-select the first loaded model (#20403)

ServeurpersoCom committed 34 days ago

Verified de190154

ggml-virtgpu: Fix some build commands (#20341)

yomaytk committed 34 days ago

Verified 05039967

metal : avoid divisions in bin kernel (#20426)

ggerganov committed 34 days ago

Verified e4cff095

ci: Setup self-hosted CI for Intel Linux Vulkan backend (#20154)

rillomas committed 34 days ago

Verified 4cc6eb15

vulkan: fix l2_norm epsilon handling (#20350)

jeffbolznv committed 34 days ago

Verified 246ffc4b

vulkan: fix OOB check in flash_attn_mask_opt (#20296)

jeffbolznv committed 34 days ago

Verified aa429cf5

vulkan: Fix ErrorOutOfHostMemory on Intel GPU when loading large models with --no-mmap (#20059)

rillomas committed 34 days ago

Verified 5866e3bb

opencl: use larger workgroup size for get_rows (#20316)

lhez committed 34 days ago

Verified 0516e04b

opencl: add cumsum op (#18981)

shaofeiqi committed 34 days ago

Verified 3d9ab225

hip: compile debug builds with -O2 on hip to avoid a compiler bug (#20392)

IMbackK committed 34 days ago

Verified d63aa398

common/parser: add GigaChatV3/3.1 models support (#19931)

Mishusha committed 34 days ago

Verified a8304b4d

model : add support for Phi4ForCausalLMV (#20168)

dranger003 committed 34 days ago

Verified fdb17643

graph : add optional scale parameter to build_lora_mm [no ci] (#20427)

richarddd committed 34 days ago

Verified 1eea6a29

common : fix --n-cpu-moe, --cpu-moe for models with fused gate + up (#20416)

ddh0 committed 34 days ago

Verified 4a748b8f

ggml-webgpu: Add supports for `GGML_OP_REPEAT` (#20230)

yomaytk committed 34 days ago

Verified f2ab047f

llama : enable chunked fused GDN path (#20340)

ggerganov committed 34 days ago

Verified d28961d8

llama : whitespace cleanup (#20422)

CISC committed 34 days ago

Verified f90bd1dd

ggml : add NVFP4 quantization type support (#19769)

richarddd committed 34 days ago

Verified 5eae9cb1

benches : add nemotron super (#20420)

ggerganov committed 34 days ago

Verified 3ca19b0e

llama : add support for Nemotron 3 Super (#20411)

danbev committed 34 days ago

Verified eaf1d793

Older