ggerganov/llama.cpp

Pull Requests Commits

gitignore : metal

ggerganov committed 2 years ago

Verified 30ac7a41

make : fix merge conflict remnants

ggerganov committed 2 years ago

Verified 28eea84a

Merge branch 'master' into build-metal-default

ggerganov committed 2 years ago

Verified 65520729

readme : update Metal instructions

ggerganov committed 2 years ago

Verified ac4038aa

common : better `n_gpu_layers` assignment

ggerganov committed 2 years ago

Verified 23360b15

ggml-opencl : store GPU buffer in ggml_tensor::extra (#2994)

slaren committed 2 years ago

Verified bd33e5ab

llama-bench : make cpp file non-executable (#2999)

cebtenzzre committed 2 years ago

Verified 31035681

make : add speculative example (#3003)

leng-yue committed 2 years ago

Verified 5b8530d8

server : add a subtle loading animation to the edit box (#2466)

akx committed 2 years ago

Verified e4386f41

2x faster (rms) norm cuda kernels (3.7% e2e improvement) (#2985)

li-plus committed 2 years ago

Verified 35195689

ggml-alloc : use virtual memory for measurement (#2973)

slaren committed 2 years ago

Verified cf9b0848

speculative : PoC for speeding-up inference via speculative sampling (#2926)

ggerganov committed 2 years ago

Verified 47068e51

perplexity : fix ETA by warming up the model with an empty run

ggerganov committed 2 years ago

Verified 8f429fa5

gguf(python): Fix special vocab handling when id < 0 (#2984)

KerfuffleV2 committed 2 years ago

Verified 6519e9c9

metal : restore 363f0bf and fix reduce in F16_F32 kernels (#2986)

ggerganov committed 2 years ago

Verified b7f2aa9e

cov : disable comment in PRs (#2989)

alonfaraj committed 2 years ago

Verified 73a12a63

llama : fix bpe tokenize from byte (#2889)

opparco committed 2 years ago

Verified 37301347

metal : revert 6af0bab until we fix it

ggerganov committed 2 years ago

Verified d9151e6f

cov : add Code Coverage and codecov.io integration (#2928)

alonfaraj committed 2 years ago

Verified afc43d5f

opencl : fix a bug in ggml_cl_pool_malloc() for ggml_cl_mul_mat_f32() (#2955)

rchardx committed 2 years ago

Verified 6460f758

metal : more optimizations (#2959)

ikawrakow committed 2 years ago

Verified ca82cf7b

llama : fix vocab_only logic when GPU is enabled

ggerganov committed 2 years ago

Verified 323a9d3b

llama : enable GPU inference by default with Metal

ggerganov committed 2 years ago

Verified 99161230

make : fix target clean

ggerganov committed 2 years ago

Verified 15f1790a

make : move targets back to the top

ggerganov committed 2 years ago

Verified b59beebd

Merge branch 'master' into build-metal-default

ggerganov committed 2 years ago

Verified 4de22829

swift : add support for k-quants (#2983)

kchro3 committed 2 years ago

Verified 6a31a3bd

convert.py : BPE fixes (#2938)

KerfuffleV2 committed 2 years ago

Verified cff7b0bf

docs : add `catai` to `README.md` (#2967)

ido-pluto committed 2 years ago

Verified 340af42f

examples : fix gpt-neox (#2943)

mmnga committed 2 years ago

Verified c42f0ec6

Older