ggerganov/llama.cpp

Pull Requests Commits

cuda : clear error after changing peer access (#10153)

slaren committed 339 days ago

Verified ea02c753

metal : simplify f16 and f32 dequant kernels (#0)

ggerganov committed 339 days ago

Verified 05697f67

metal : move dequantize templates to beginning of MSL source (#0)

ggerganov committed 339 days ago

Verified f8e58135

CANN: adjust backend registry refactor. (#10158)

leo-pony committed 339 days ago

Verified 329ed914

ggerganov committed 339 days ago

Verified ce027adf

cmake : make it possible linking ggml as external lib (ggml/1003)

ykhrustalev committed 339 days ago

Verified 284e5b02

metal : fix minor string leaks (ggml/1004)

pminev committed 339 days ago

Verified e2292aaa

ggml : move CPU backend to a separate file (#10144)

slaren committed 340 days ago

Verified 9f409893

metal : minor fixup in FA kernel (#10143)

ggerganov committed 340 days ago

Verified 08828a6d

flake.lock: Update (#10146)

ggerganov committed 340 days ago

Verified 1839f691

Add apple arm to presets (#10134)

kohnech committed 341 days ago

Verified 9830b692

server : fix slot selection by lru (#10126)

sasha0552 committed 341 days ago

Verified 42cadc74

server : fix endpoint checks (#10135)

ggerganov committed 341 days ago

Verified 45950415

llama : adjust default context size + print warnings (#10136)

ggerganov committed 341 days ago

Verified 1926d6e3

simple-chat : only add bos on first prompt (#10129)

slaren committed 341 days ago

Verified b634f8a2

convert-lora : make `--base` optional (#10110)

ngxson committed 341 days ago

Verified 7554aa46

llama : add simple-chat example (#10124)

slaren committed 342 days ago

Verified a6744e43

llama : use smart pointers for ggml resources (#10117)

slaren committed 342 days ago

Verified e991e312

vulkan : improve ggml_vk_create_buffer error handling (#9898)

FanShupei committed 342 days ago

Verified 418f5eef

readme : update hot topics

ggerganov committed 342 days ago

Verified ba6f62eb

server : fix smart selection of available slot (#10120)

sasha0552 committed 342 days ago

Verified d865d147

ggml : remove ggml_scratch (#10121)

ggerganov committed 342 days ago

Verified 1804adb0

ggerganov committed 342 days ago

Verified 815fe72a

ggml : alloc ggml_contexts on the heap (whisper/2525)

ggerganov committed 342 days ago

Verified f221d562

build: fix build error in Windows env with OneAPI setup (#10107)

kylo5aby committed 343 days ago

Verified e597e507

llama : improve output buffer type selection (#10098)

slaren committed 343 days ago

Verified 85679d37

quantize : fix --keep-split (#10114)

slaren committed 343 days ago

Verified 1e9f9499

llama : fix buffer checks for mamba and rwk (#10111)

slaren committed 343 days ago

Verified c02e5ab2

loader: refactor tensor weights storage (#9935)

kylo5aby committed 343 days ago

Verified ab3d71f9

server : include scheme when printing URL (#10106)

bakkot committed 343 days ago

Verified 0a683e80

Newer Older