ggerganov/llama.cpp

Pull Requests Commits

ggml : fix 32-bit ARM compat (cont)

ggerganov committed 1 year ago

Verified 7216af5c

ggml : fix vld1q_s8_x4 32-bit compat

ggerganov committed 1 year ago

Verified 27afe299

common : fix the short form of `--grp-attn-w`, not `-gat` (#4825)

howlger committed 1 year ago

Verified 1fc2f265

readme : add link to SOTA models

ggerganov committed 1 year ago

Verified a9a8c5de

SOTA 2-bit quants (#4773)

ikawrakow committed 1 year ago

Verified dd5ae064

swift : exclude ggml-metal.metal from the package (#4822)

ggerganov committed 1 year ago

Verified 668b31fc

llama.swiftui : update readme

ggerganov committed 1 year ago

Verified 42ea63c5

main : add self-extend support (#4815)

ggerganov committed 1 year ago

Verified 52531fdf

examples : add passkey test (#3856)

ggerganov committed 1 year ago

Verified b0034d93

readme : add lgrammel/modelfusion JS/TS client for llama.cpp (#4814)

lgrammel committed 1 year ago

Verified b7e79829

llama-bench : add no-kv-offload parameter (#4812)

slaren committed 1 year ago

Verified 226460cc

CUDA: fixed redundant value dequantization (#4809)

JohannesGaessler committed 1 year ago

Verified d5a410e8

llama : remove unused vars (#4796)

ggerganov committed 1 year ago

Verified 9dede37d

llama : remove redundant GQA check (#4796)

ggerganov committed 1 year ago

Verified 3c36213d

llama.swiftui : use llama.cpp as SPM package (#4804)

azarovalex committed 1 year ago

Verified 72d8407b

llama : print tensor meta for debugging

ggerganov committed 1 year ago

Verified d117d4dc

llama.swiftui : add visionOS target (#4805)

azarovalex committed 1 year ago

Verified 3418c03e

ggml : use __builtin_amdgcn_sudot4 in __dp4a for gfx11 (#4787)

kzhuravl committed 1 year ago

Verified 63ee677e

server : fix n_predict check (#4798)

ggerganov committed 1 year ago

Verified 67984921

llama.swiftui : use correct pointer for llama_token_eos (#4797)

illescasDaniel committed 1 year ago

Verified c75ca5d9

examples : improve base-translate.sh script (#4783)

ggerganov committed 1 year ago

Verified 96e80dab

cmake : check for openblas64 (#4134)

a-n-n-a-l-e-e committed 1 year ago

Verified eec22a1c

flake.nix : fix typo (#4700)

eltociear committed 1 year ago

Verified be36bb94

metal : switch back to default.metallib (ggml/681)

ggerganov committed 1 year ago

91d38876

ggml : fix q2_k bpw in comments (ggml/680)

ggerganov committed 1 year ago

d061bf94

ggml : add error handling to graph_compute (whisper/1714)

finnvoor committed 1 year ago

1bf681f9

ggml : do not sched_yield when calling BLAS (#4761)

ggerganov committed 1 year ago

Verified c1d7cb28

examples : add few-shot translation example (#4783)

ggerganov committed 1 year ago

Verified 3681f224

finetune : remove unused includes (#4756)

danbev committed 1 year ago

Verified b3a7c20b

server : send token probs for "stream == false" (#4714)

ggerganov committed 1 year ago

Verified 012cf349

Older