ggerganov/llama.cpp

Pull Requests Commits

YaRN : correction to GPT-NeoX implementation

cebtenzzre committed 1 year ago

f8249026

llama : restore prefix space in llama tokenizer (#4081)

cebtenzzre committed 1 year ago

Verified a6fc554e

ggml-cuda : increase max graph size (#4084)

slaren committed 1 year ago

Verified 1cf2850d

Fix MacOS Sonoma model quantization (#4052)

TortoiseHam committed 1 year ago

Verified 6bb4908a

stablelm : StableLM support (#3586)

Galunid committed 1 year ago

Verified 36eed0c4

convert.py: also look for plain model.safetensors (#4043)

afrideva committed 1 year ago

Verified b46d12f8

llava : fix regression for square images in #3613 (#4056)

monatis committed 1 year ago

Verified bd90eca2

ggml : sync (im2col, GPU conv, 32-bit arm compat) (#4060)

ggerganov committed 1 year ago

Verified 3d68f364

readme : update hot topics

ggerganov committed 1 year ago

Verified c049b37d

sync : ggml (backend v2) (#3912)

ggerganov committed 1 year ago

Verified 4760e7cc

Add ReLU and SQR CUDA ops to (partially) fix Persimmon offloading (#4041)

KerfuffleV2 committed 1 year ago

Verified bb50a792

gguf-py: gguf_writer: Use bytearray to build metadata (#4051)

KerfuffleV2 committed 1 year ago

Verified 21fd874c

Fix some documentation typos/grammar mistakes (#4032)

richardkiss committed 1 year ago

Verified 532dd74e

Fix gguf-convert-endian script (#4037)

monatis committed 1 year ago

Verified e86fc56f

server : fix crash when prompt exceeds context size (#3996)

z80maniac committed 1 year ago

Verified d96ca7de

gguf-py: Refactor and allow reading/modifying existing GGUF files (#3981)

KerfuffleV2 committed 1 year ago

Verified 34b0a082

server : allow continue edit on completion mode (#3950)

jhen0409 committed 1 year ago

Verified 4a4fd3ee

Unbreak persimmon after #3837 (#4010)

Galunid committed 1 year ago

Verified df9d1293

scripts: Generalize convert scripts (#3838)

Galunid committed 1 year ago

Verified a75fa576

server : add min_p param (#3877)

Mihaiii committed 1 year ago

Verified 57ad015d

ggml-alloc : fix backend assignments of views (#3982)

slaren committed 1 year ago

Verified 875fb428

gguf : track writer state, free unneeded tensors, cleanup (#3871)

cebtenzzre committed 1 year ago

Verified 0a7c980b

make : do not add linker flags when compiling static llava lib (#3977)

ggerganov committed 1 year ago

Verified 413503d4

ggml : fix backward rope after YaRN (#3974)

xaedes committed 1 year ago

Verified e9c1cecb

Use params when loading models in llava-cli (#3976)

tejom committed 1 year ago

Verified 54b4df88

cuda : supports running on CPU for GGML_USE_CUBLAS=ON build (#3946)

wsxiaoys committed 1 year ago

Verified 46876d2a

llava : expose as a shared library for downstream projects (#3613)

damian0815 committed 1 year ago

Verified 381efbf4

ggml-cuda : fix f16 mul mat (#3961)

slaren committed 1 year ago

Verified 2833a6f6

Allow common process_escapes to handle \x sequences (#3928)

KerfuffleV2 committed 1 year ago

Verified d9ccce2e

server : fix typo for --alias shortcut from -m to -a (#3958)

RoyalHeart committed 1 year ago

Verified bb60fd0b

Older