ggerganov/llama.cpp

Pull Requests Commits

llama : add SKIP_KQ_KQV option

ggerganov committed 1 year ago

Verified cb79f8a2

ggml : skip nops

ggerganov committed 1 year ago

Verified ed9fde7a

llama : profiling the attention compute

ggerganov committed 1 year ago

Verified 2471d56a

batched : add len CLI argument

ggerganov committed 1 year ago

Verified 22c69a27

CLBlast: Add outer loops over src0 for broadcasting in mulmat

shibe2 committed 1 year ago

465219b9

sampling : refactor init to use llama_sampling_params (#3696)

ggerganov committed 1 year ago

Verified d1031cf4

gguf : support big endian platform (#3552)

chenqiny committed 1 year ago

Verified 8cf19d60

server : fix uninitialized sampling context (close #3685)

ggerganov committed 1 year ago

Verified a0edf73b

ggml : fix rope + llama minor optimizations (#3560)

GermanAizek committed 1 year ago

Verified f439e506

convert : restore compat with old Falcon models (#3680)

cebtenzzre committed 1 year ago

Verified e78f3ef2

multimodal : add BakLLaVA conversion support (#3682)

monatis committed 1 year ago

Verified f3b25e40

llava : avoid segfault in case of non-existent mmproj file (#3674)

monatis committed 1 year ago

Verified 60abea97

readme : update hot topics

ggerganov committed 1 year ago

Verified 004797f6

speculative : bug fixes

ggerganov committed 1 year ago

Verified 4e82b2ea

speculative : add tree-based sampling example (#3624)

ggerganov committed 1 year ago

Verified 0e89203b

metal : implement q5_0 and q5_1 kernels (#3648)

jhen0409 committed 1 year ago

Verified c67fe68e

opencl : fix element-wise multiplication (#3656)

shibe2 committed 1 year ago

Verified 1117d066

fix embeddings when using CUDA (#3657)

slaren committed 1 year ago

Verified cb33f43a

llama : avoid fprintf in favor of LLAMA_LOG (#3538)

ggerganov committed 1 year ago

Verified e1675d13

readme : update hot-topics & models, detail windows release in usage (#3615)

BarfingLemurs committed 1 year ago

Verified 8402566a

CLBlast: Fix temporary buffer size for f16 conversion (wsize)

shibe2 committed 1 year ago

40e5ce05

train-text-from-scratch : fix assert failure in ggml-alloc (#3618)

slaren committed 1 year ago

Verified a5e8c1d8

editorconfig : remove trailing spaces

ggerganov committed 1 year ago

Verified e74c705e

server : documentation of JSON return value of /completion endpoint (#3632)

coezbek committed 1 year ago

Verified 3ad1e3f1

save-load-state : fix example + add ci test (#3655)

ggerganov committed 1 year ago

Verified 1142013d

readme : add Aquila2 links (#3610)

ftgreat committed 1 year ago

Verified 5fe268a4

tokenizer : special token handling (#3538)

staviq committed 1 year ago

Verified 1a159553

k-quants : fix quantization ranges (#3646)

ggerganov committed 1 year ago

Verified 281ef73c

llava : fix tokenization to not add bos between image embeddings and user prompt (#3645)

ggerganov committed 1 year ago

Verified 940efa95

MPT : support GQA for replit-code-v1.5 (#3627)

cebtenzzre committed 1 year ago

Verified 11bff290

Older