ngxson/llama.cpp

Pull Requests Commits

CUDA: fix MMQ writeback for int8 tensor cores (#8100)

JohannesGaessler committed 1 year ago

Verified 3b099bcd

CUDA: use MMQ instead of cuBLAS by default (#8075)

JohannesGaessler committed 1 year ago

Verified a818f302

gguf-py : fix tensor groups for encoder-decoder models in gguf-dump.py (#8090)

fairydreaming committed 1 year ago

Verified d62e4aaa

CUDA: optimize MMQ int8 tensor core performance (#8062)

JohannesGaessler committed 1 year ago

Verified 9a590c82

Option to split during conversion (#6942)

christianazinn committed 1 year ago

Verified 52fc8705

disable publishing the full-rocm docker image (#8083)

slaren committed 1 year ago

Verified 8cb508d0

embedding : more cli arguments (#7458)

YannFollet committed 1 year ago

Verified 646ef4a9

gguf-py, convert-hf : model conversion support for T5 and FLAN-T5 model variants (#5763)

fairydreaming committed 1 year ago

Verified de0d6a68

ggml : remove ggml_task_type and GGML_PERF (#8017)

slaren committed 1 year ago

Verified 95f57bb5

llama : add support for BitnetForCausalLM (#7931)

Eddie-Wang1120 committed 1 year ago

Verified e112b610

server : fix JSON-Scheme typo (#7975)

akx committed 1 year ago

Verified 6a2f298b

Fix typo in llama_set_embeddings comment (#8077)

danbev committed 1 year ago

Verified 11318d9a

fix CI failures (#8066)

slaren committed 1 year ago

Verified b6b9a8e6

Refactor Vulkan backend to allow multiple contexts (#7961)

0cc4m committed 1 year ago

Verified 45c0e2e4

Removing extra blank lines that were breaking Lint. (#8067)

HanClinto committed 1 year ago

Verified b5a5f34e

cvector: fix CI + correct help message (#8064)

ngxson committed 1 year ago

Verified 3e58b0ee

cvector-generator: Moe Moe Fixie-Fixie for Lots of Formats~! ♡(ᐢ ᴥ ᐢ)♡ (#8052)

HatsuneMikuUwU33 committed 1 year ago

Verified adf480c3

convert-hf : change assert to exception (#8015)

0xspringtime committed 1 year ago

Verified 3aa184a8

Update llama-quantize ppl/file size output from LLaMA-v1 to Llama-3 values (#8058)

ddh0 committed 1 year ago

Verified 5b48cd53

JSON Schema to GBNF integration tests (#7790)

HanClinto committed 1 year ago

Verified c5a8d4b7

vulkan: detect multiple devices by deviceUUID instead of deviceID (#8022)

Adriankhl committed 1 year ago

Verified 557b653d

ggml : AVX IQ quants (#7845)

netrunnereve committed 1 year ago

Verified 7d5e8777

llama : optimize long word tokenization with WPM (#8034)

ggerganov committed 1 year ago

Verified a927b0f3

llama : allow pooled embeddings on any model (#7477)

iamlemec committed 1 year ago

Verified 80ea089d

swiftui : enable stream updating (#7754)

shu223 committed 1 year ago

Verified 0e64591e

requirements : Bump torch and numpy for python3.12 (#8041)

hamdoudhakem committed 1 year ago

Verified b1ef562b

convert-hf : Fix the encoding in the convert-hf-to-gguf-update.py (#8040)

hamdoudhakem committed 1 year ago

Verified 17b291a6

common: fix warning (#8036)

JohannesGaessler committed 1 year ago

Verified abd894ad

[SYCL] Fix windows build and inference (#8003)

luoyu-intel committed 1 year ago

Verified de391e4c

CUDA: stream-k decomposition for MMQ (#8018)

JohannesGaessler committed 1 year ago

Verified d50f8897

Newer Older