ngxson/llama.cpp

Pull Requests Commits

mtmd : clean up clip_n_output_tokens

ngxson committed 126 days ago

e9ef96af

ggml-quants : fix make_qp_quants NANs and IQ1 assertion errors (#15379)

compilade committed 126 days ago

Verified f44f7931

vulkan: disable spirv-opt for bfloat16 shaders (#15352)

jeffbolznv committed 126 days ago

Verified ae532eac

server : export max observed n_past value (#15361)

okuvshynov committed 126 days ago

Verified e5155e69

vulkan: Use larger workgroups for mul_mat_vec when M is small (#15355)

jeffbolznv committed 127 days ago

Verified 21c17b5b

vulkan: support sqrt (#15370)

ddwkim committed 127 days ago

Verified 19f4deca

convert : force patch_embd weights to F16 or F32 to avoid broken GGUFs (#15367)

CISC committed 127 days ago

Verified 4d196981

ci : fix hang in windows-hip build/release (#15365)

CISC committed 127 days ago

Verified b143fbc8

vulkan: Optimize argsort (#15354)

jeffbolznv committed 127 days ago

Verified de562791

model : support vision LiquidAI LFM2-VL family (#15347)

tdakhran committed 127 days ago

Verified 65349f26

vulkan: fuse adds (#15252)

jeffbolznv committed 127 days ago

Verified 1fe00296

vulkan: Support mul_mat_id with f32 accumulators (#15337)

jeffbolznv committed 128 days ago

Verified de219279

vulkan: Add missing bounds checking to scalar/coopmat1 mul_mat_id (#15334)

jeffbolznv committed 128 days ago

Verified 2e2b22ba

OpenCL: add initial FA support (#14987)

rmatif committed 128 days ago

Verified 912ff8c1

common : fix double bos, use common_chat_templates for add_bos and add_eos (#15326)

danbev committed 128 days ago

Verified 5e6229a8

opencl: add initial mxfp4 support via mv (#15270)

lhez committed 128 days ago

Verified e2c1bfff

vulkan : fix out-of-bounds access in argmax kernel (#15342)

ggerganov committed 129 days ago

Verified 5edf1592

vulkan : fix compile warnings on macos (#15340)

ggerganov committed 129 days ago

Verified db3010bd

ggml: initial IBM zDNN backend (#14975)

taronaeo committed 129 days ago

Verified ff27f80a

ci : fix ios-xcode-build (#15324)

CISC committed 129 days ago

Verified d3248d9b

ci : move ccache action to ggml-org fork (#15328)

slaren committed 129 days ago

Verified 7aeee88c

test-opt: fix backend support check (#15317)

JohannesGaessler committed 129 days ago

Verified b07791aa

CUDA: fix negative KV_max values in FA (#15321)

JohannesGaessler committed 129 days ago

Verified 4227c9be

eval-callback : stop on first NaN (#15320)

ggerganov committed 129 days ago

Verified df36bce6

chat : include kwargs in template example (#15309)

slaren committed 129 days ago

Verified f75b8306

llama : add 18-layer model type for Gemma 3-270m (#15319)

danbev committed 130 days ago

Verified 7a0de960

devops : fix compile bug when the BASE_CUDA_DEV_CONTAINER is based on Ubuntu 24.04 (#15005)

simevo committed 130 days ago

Verified e4e91591

HIP: Cleanup hipification header (#15285)

IMbackK committed 130 days ago

Verified 5ba36f61

gpt-oss: implement harmony parsing (#15181)

aldehir committed 130 days ago

Verified b204a5a2

docker : Enable GGML_CPU_ALL_VARIANTS for ARM (#15267)

ckastner committed 130 days ago

Verified 646944cf

Older