ggerganov/llama.cpp

Pull Requests Commits

hf : add support for --repo and --file

ggerganov committed 2 years ago

Verified e856bfed

hf : add error logs

ggerganov committed 2 years ago

Verified e834aa1f

scripts : add hf.sh helper scripts

ggerganov committed 2 years ago

Verified 303da634

vulkan: Find optimal memory type but with fallback (#5381)

luciferous committed 2 years ago

Verified 704359e2

readme : fix typo (#5490)

Rune-AI committed 2 years ago

Verified 594fca3f

llava : update README.md (#5489)

cmp-nct committed 2 years ago

Verified ccbb277f

cmake : ARM intrinsics detection for MSVC (#5401)

Xarbirus committed 2 years ago

Verified 8084d554

llava : support v1.6 (#5267)

cmp-nct committed 2 years ago

Verified aa234129

Early return for zero size calls to get_tensor. (#5482)

manyoso committed 2 years ago

Verified f5ca0548

gguf : add python reader example (#5216)

cmp-nct committed 2 years ago

Verified 6c00a066

llama : add support for Nomic Embed (#5468)

cebtenzzre committed 2 years ago

Verified ea9c8e11

llama : allow raw byte in SPM vocabs; don't crash on nl 404 (#5478)

akx committed 2 years ago

Verified c4e6dd59

llama : make load error reporting more granular (#5477)

akx committed 2 years ago

Verified 037259be

finetune : rename feed-forward tensors (w1/w2/w3) (#4839)

danbev committed 2 years ago

Verified 26397890

tests : multi-thread the tokenizer tests (#5474)

ggerganov committed 2 years ago

Verified cf45252a

llama : support batched embeddings (#5466)

iamlemec committed 2 years ago

Verified 03bf161e

make: add error message for bad CUDA version (#5444)

JohannesGaessler committed 2 years ago

Verified ad014bba

bert : add tests + fix quantization (#5475)

ggerganov committed 2 years ago

Verified 49cc1f7d

tests : disable moe test (#5473)

ggerganov committed 2 years ago

Verified 99b8b43d

ggml-quants : fix compiler warnings (shadow variable) (#5472)

ikawrakow committed 2 years ago

Verified 895407f3

llama : fix quantization when tensors are missing (#5423)

ggerganov committed 2 years ago

Verified 099afc62

swift : package no longer use ggml dependency (#5465)

ggerganov committed 2 years ago

Verified df334a11

py : fix persimmon `n_rot` conversion (#5460)

lx200916 committed 2 years ago

Verified dbd8828e

ggml-sycl: Replace 3d ops with macro (#5458)

abhilash1910 committed 2 years ago

Verified 43fe07c1

llava : remove prog parameter from ArgumentParser (#5457)

danbev committed 2 years ago

Verified 4a46d2b7

sync : ggml (#5452)

ggerganov committed 2 years ago

Verified 3b169441

CUDA: mul_mat_vec_q tiling, refactor mul mat logic (#5434)

JohannesGaessler committed 2 years ago

Verified 3bdc4cd0

Add support for BERT embedding models (#5423)

iamlemec committed 2 years ago

Verified 2891c8aa

flake.lock: Update

github-actions[bot] committed 2 years ago

97a33650

vulkan: only use M-sized matmul on Apple GPUs (#5412)

slp committed 2 years ago

Verified c88c74f9

Older