ggerganov/llama.cpp

Pull Requests Commits

llama : fix incorrect Qwen2Moe ffn_moe_out graph callback (#12631)

CISC committed 224 days ago

Verified 3714c3ee

metal : improve FA + improve MoE (#12612)

ggerganov committed 225 days ago

Verified b4ae5081

vulkan: fix coopmat shader generation when cross-compiling (#12272)

Icenowy committed 225 days ago

Verified b86f6007

llama: fix error on bad grammar (#12628)

JohannesGaessler committed 225 days ago

Verified dd373dd3

server : include speculative decoding stats when timings_per_token is enabled (#12603)

mostlygeek committed 225 days ago

Verified 5d016702

rpc : update README for cache usage (#12620)

rgerganov committed 225 days ago

Verified ef03229f

llamafile : ppc64le GEMV forwarding for FP32. (#12594)

amritahs-ibm committed 225 days ago

Verified 13731766

rpc : send hash when tensor data is above some fixed threshold (#12496)

rgerganov committed 225 days ago

Verified ab6ab8f8

server : Support listening on a unix socket (#12613)

p1-0tr committed 225 days ago

Verified 2099a9d5

media : add SVG logo [no ci] (#12616)

ggerganov committed 225 days ago

Verified 29690198

opencl: add multi and vision rope, `gelu_quick` and `im2col` (#12600)

lhez committed 226 days ago

Verified 5dec47dc

llama : add PLM GGUF Conversion & Inference Support (#12457)

Si1w committed 226 days ago

Verified f125b8dc

model : restore support for T5Encoder (#12590)

HighDoping committed 226 days ago

Verified 953c2a62

convert : Support Qwen2_5_VLForConditionalGeneration (#12595)

csabakecskemeti committed 226 days ago

Verified d5c6309d

ggerganov committed 226 days ago

029c693f

scripts : update sync + fix cmake merge

ggerganov committed 226 days ago

771d8437

ggerganov committed 226 days ago

df0665a4

cmake : sync/merge PowerPC build commands (#0)

ggerganov committed 226 days ago

0306aad1

llamafile : ppc64le MMA implementation for Q4_0. (#12489)

amritahs-ibm committed 226 days ago

Verified c7b43ab6

ggml : riscv: add 128-bit RVV support (#12530)

xctan committed 226 days ago

Verified 24feaec0

llama : make loras compatible with repacking (#12593)

ggerganov committed 226 days ago

Verified f28bc4c2

SYCL: implement memset ggml backend buffer interface (#12580)

qnixsynapse committed 226 days ago

Verified f17a3bb4

HIP: Add support for RDNA4 targets (#12372)

slojosic-amd committed 226 days ago

Verified bd40678d

metal : refactor mat-vec code (#12569)

ggerganov committed 227 days ago

Verified b3298fa4

upgrade to llguidance 0.7.10 (#12576)

mmoskal committed 227 days ago

Verified 2447ad8a

clip: Fix llama-llava-clip-quantize-cli quantization error under CUDA backend (#12566)

Ivy233 committed 227 days ago

Verified 02082f15

convert : fix squeeze for ssm_conv tensors (#12573)

ggerganov committed 227 days ago

Verified df4d20cd

ggml : fix MUL_MAT_ID repack with Q8_K (#12544)

ggerganov committed 227 days ago

Verified 5ed38b68

doc: [MUSA] minor changes (#12583)

yeahdongcn committed 227 days ago

Verified fd7855f8

convert: fix Mistral3/Gemma3 model hparams init (#12571)

CISC committed 227 days ago

Verified 53af4dba

Newer Older