ggerganov/llama.cpp

Pull Requests Commits

Add mmap pages stats (disabled by default)

prusnak committed 2 years ago

Verified 15067374

stdout : vertical align outputs for better readibility

ggerganov committed 2 years ago

Verified 3173a62e

examples: add missing <ctime> include for time() (#1011)

prusnak committed 2 years ago

Verified 489537e6

Fix msys2 build error and warnings (#1009)

na-na-hi committed 2 years ago

Verified 2d3481c7

convert.py: Fix loading safetensors and ggml format on Windows (#991)

comex committed 2 years ago

Verified 74f5899d

Fix potential int8 overflow in non-SIMD vec_dot (#986)

sw committed 2 years ago

Verified 2f7c8e01

Refactor ggml.c for future tensor types (#1001)

sw committed 2 years ago

Verified 0ad96463

ggml : add Q8_0 quantization for intermediate results (#951)

ggerganov committed 2 years ago

Verified e95b6554

ggml : use posix_memalign on non-Windows env

ggerganov committed 2 years ago

Verified aa485cee

benchmark : fix result validation in benchmark-q4_0-matmult (#987)

dfyz committed 2 years ago

Verified c12b14b7

cmake : add finding the OpenBLAS header file (#992)

katsu560 committed 2 years ago

Verified 106faaf2

Revert "main : alternative instruct mode (Vicuna support, etc.) (#863)" (#982)

prusnak committed 2 years ago

Verified c85e03d1

py : bump sentencepiece to 0.1.98 to support Python 3.11 (#976)

prusnak committed 2 years ago

Verified 48909354

make : fix dependencies, use auto variables (#983)

sw committed 2 years ago

Verified 93265e98

Expose type name from ggml (#970)

prusnak committed 2 years ago

Verified c56b7152

main : alternative instruct mode (Vicuna support, etc.) (#863)

aroidzap committed 2 years ago

Verified f4d277ae

ggml : add unary and binary map operations (#874)

KerfuffleV2 committed 2 years ago

Verified c9a59b70

py : cleanup dependencies (#962)

prusnak committed 2 years ago

Verified a32f7acc

py : fix flake8 and isort nitpicks (#960)

prusnak committed 2 years ago

Verified 43ffdefb

ggerganov committed 2 years ago

Verified 1623a6e9

ggml : always allocate buffers with size multiple of GGML_MEM_ALIGN

ggerganov committed 2 years ago

Verified c14e0d2f

py : new conversion script (#545)

comex committed 2 years ago

Verified 723dac55

ggml : fix q4_1 dot product types

ggerganov committed 2 years ago

Verified 0f07cacb

ggml : optimize rope function to avoid call powf in the tight loop (#807)

howard0su committed 2 years ago

Verified c5d70f5c

perplexity : add support for batch size to `--perplexity` (#407)

glinscott committed 2 years ago

Verified be87b6ed

common : remove unnecessary includes (#947)

CRD716 committed 2 years ago

Verified 0e07e6a8

ggml : add GGML_DEFAULT_N_THREADS

ggerganov committed 2 years ago

Verified a3a2a0ed

ggml : speed-up ggml_vec_dot_q4_1() ARM_NEON + 32-bit ARM support (#900)

ggerganov committed 2 years ago

Verified d990e3ff

llama : merge llama_internal.h into llama.h

ggerganov committed 2 years ago

Verified 9190e8ea

gitignore : benchmark

ggerganov committed 2 years ago

Verified c85980ac

Older