ngxson/llama.cpp

Pull Requests Commits

server: poc audio gen

ngxson committed 81 days ago

37bf306b

vulkan: reject ops when a tensor is too large to allocate (#18646)

jeffbolznv committed 81 days ago

Verified ca4a8370

vulkan: Warptile tuning for Intel Xe2/Xe3 (#18178)

virajwad committed 81 days ago

Verified 03023296

vulkan: more mul mat optimizations (#18533)

netrunnereve committed 81 days ago

Verified 8c77a04c

examples : add debug utility/example (#18464)

danbev committed 81 days ago

Verified ffba4f29

CANN: Fix rename for get_env (#18652)

hipudding committed 82 days ago

Verified 3333951d

CANN: Rename `get_env` to `get_env_as_lowercase` (#18624)

rauletorresc committed 82 days ago

Verified 193ee38a

Hexagon add support for f16/f32 flash attention, scale, set-rows and improve f16/32 matmul (#18611)

max-krasnyansky committed 82 days ago

Verified 95ea9e08

mtmd: mtmd_audio_streaming_istft (#18645)

tdakhran committed 82 days ago

Verified ccbc84a5

llama-params-fit: fix last devices with low VRAM (#18494)

JohannesGaessler committed 82 days ago

Verified 68b4d516

ggml : optimize cuda ssm_scan using warp-level reduction (#18505)

Aadeshveer committed 82 days ago

Verified 24af22fc

arg: use CSV escape style for multiple-value args (#18643)

ngxson committed 82 days ago

Verified 07fbe19f

vulkan: support buffer_from_host_ptr (#18467)

jeffbolznv committed 82 days ago

Verified ea13cba8

ggml-cuda: refactor cuda graph usage (#18637)

am17an committed 82 days ago

Verified 090b137e

mmq.cu: tune mmq/rocblas switching for RDNA (#18537)

Beinsezii committed 82 days ago

Verified 96892952

server : add thinking content blocks to Anthropic Messages API (#18551)

686f6c61 committed 82 days ago

Verified 3d26a09d

gguf-py : add requests to dependencies (#18629)

ckastner committed 83 days ago

Verified bd2a93d4

ggml : fix avx512bf16 build (#18623)

angt committed 83 days ago

Verified e75ee110

CANN: Make `valid_values` variable `static const` (#18627)

rauletorresc committed 83 days ago

Verified da9b8d33

ggml webgpu: add CEIL operation support (#18605)

nwyin committed 83 days ago

Verified e443fbcf

model : add LFM2-ColBert-350M (#18607)

tdakhran committed 83 days ago

Verified 73d284a2

CUDA: fix FA FP16 accumulator overflow for Granite (#18614)

JohannesGaessler committed 83 days ago

Verified df17a4c9

add YoutuVLForConditionalGeneration architectures (#18620)

f291400 committed 83 days ago

Verified 1871f0ba

ggml-cuda: check for srcs outside the cgraph (#18583)

am17an committed 83 days ago

Verified f47edb8c

server : fix router child env in containerized environments (#18562)

v-sayapin committed 83 days ago

Verified da143b99

vulkan: fix topk_moe_sigmoid_norm_bias failures in GLM-4.6 (#18582)

jeffbolznv committed 83 days ago

Verified f1768d8f

models : fix backend assignment for Granite/Nemotron graphs (#18599)

ggerganov committed 83 days ago

Verified 2da64a2f

vulkan: handle quantize_q8_1 overflowing the max workgroup count (#18515)

jeffbolznv committed 83 days ago

Verified b37124d2

llama : refactor rope_freq_base/scale_swa conversion and init (#18553)

CISC committed 84 days ago

Verified eadc4184

CANN: add operator fusion support for ADD + RMS_NORM (#17512)

noemotiovon committed 84 days ago

Verified 67e3f6f6

Older