ggml-org/ggml

Pull Requests Commits

files : remove old sources

ggerganov committed 1 year ago

Verified c29a3a3e

sync : llama.cpp

ggerganov committed 1 year ago

c9376dcf

metal : use less stack memory in FA kernel (llama/14088)

ggerganov committed 1 year ago

c4f77915

ggml-cpu : split arch-specific implementations (llama/13892)

xctan committed 1 year ago

0aef4bcf

cuda : fix device sync on buffer clear (llama/14033)

slaren committed 1 year ago

fd175573

CANN: Simplify the environment variable setting(#13104)

bachelor-dou committed 1 year ago

8ca45113

sycl: Add reorder to Q6_K mmvq implementation (llama/13885)

s-Nick committed 1 year ago

2640ba54

cuda : fix buffer type check with integrated GPUs (llama/14069)

slaren committed 1 year ago

22c81e2d

SYCL: Implement few same quantized type copy kernels (llama/13739)

qnixsynapse committed 1 year ago

b42f95e3

vulkan: Enable VK_KHR_cooperative_matrix extension for Intel Xe2 GPUs (llama/14001)

rillomas committed 1 year ago

852e354d

llama : allow using mmap without PrefetchVirtualMemory, apply GGML_WIN_VER to llama.cpp sources (llama/14013)

slaren committed 1 year ago

92110104

vulkan: automatically deduce size of push constants (llama/13936)

jeffbolznv committed 1 year ago

d215fedb

ggml-vulkan: adds support for op CONV_TRANSPOSE_1D (llama/13813)

etasnadi committed 1 year ago

cb9b2945

releases : use dl backend for linux release, remove arm64 linux release (llama/13996)

slaren committed 1 year ago

e9946fa3

CUDA: fix FTZ in FA for Gemma 3 (llama/13991)

JohannesGaessler committed 1 year ago

d935d466

vulkan: fix warnings in perf logger querypool code (llama/13937)

jeffbolznv committed 1 year ago

bf6c5aa3

opencl: add `backend_synchronize` (llama/13939)

lhez committed 1 year ago

6e8b80de

OpenCL: Add concat, tsembd, upscale, tanh, pad and repeat (llama/13840)

rmatif committed 1 year ago

6bc3202d

metal : use F32 accumulators in FA kernels (llama/13975)

ggerganov committed 1 year ago

8bb8dbca

cmake : Handle mixed-case 'Power' strings in POWER CPU detection (llama/13966)

shalinib-ibm committed 1 year ago

85d39e2e

sycl: quantize and reorder the input to q8_1 when reorder is enabled (llama/13826)

AD2605 committed 1 year ago

dc0b52e6

gguf: fix failure on version == 0 (llama/13956)

JohannesGaessler committed 1 year ago

dbbbe0c2

ggml: check if non-native endian model is being loaded (llama/13943)

taronaeo committed 1 year ago

3a9ce1fb

sam : add default case for unsupported prompt types (#1263)

danbev committed 1 year ago

Verified a1761cd6

Add in-build ggml::ggml ALIAS library (#1260)

dg0yt committed 1 year ago

Verified 988abe2a

yolo : add support for dl backends (#1251)

rgerganov committed 1 year ago

Verified a61135c6

sync : whisper.cpp

ggerganov committed 1 year ago

3185d495

sync : llama.cpp

ggerganov committed 1 year ago

adeb7a04

threading: support for GGML_SCHED_PRIO_LOW, update thread info on Windows to avoid throttling (llama/12995)

max-krasnyansky committed 1 year ago

bb57c398

CUDA: add a prop in ggml_cuda_device_infor for distinguish iGPU or dGPU in cuda (#13856) (llama/13895)

Yangxiaoz committed 1 year ago

078a221f

Older