ggerganov/llama.cpp

Pull Requests Commits
metal : correctly set SIMD support flags on iOS (#4923)
azarovalex committed 1 year ago
Verified 5f5fe1bd
llama : support WinXP build with MinGW 8.1.0 (#3419)
guilt committed 1 year ago
Verified ac32902a
2-bit quantizations (#4897)
ikawrakow committed 1 year ago
Verified 147b17ac
Make Q3_K_S be the same as olf Q3_K_L for Mixtral-8x7B (#4906)
ikawrakow committed 1 year ago
Verified 807179ec
sync : ggml
ggerganov committed 1 year ago
Verified 76484fbf
ggml: cache sin/cos for RoPE (#4908)
JohannesGaessler committed 1 year ago
Verified c71d608c
metal : remove old API (#4919)
ggerganov committed 1 year ago
Verified 4be5ef55
server : fix prompt caching with system prompt (#4914)
ggerganov committed 1 year ago
Verified 0ea069b8
llama : fix detokenization of non-special added-tokens (#4916)
ggerganov committed 1 year ago
Verified f172de03
metal : disable log for loaded kernels (#4794)
ggerganov committed 1 year ago
Verified 2d57de52
llama : minimize size used for state save/load (#4820)
dfriehs committed 1 year ago
Verified df845cc9
workflows: unbreak nix-build-aarch64, and split it out (#4915)
SomeoneSerge committed 1 year ago
Verified 6b48ed08
main : add parameter --no-display-prompt (#4541)
YannFollet committed 1 year ago
Verified 722d33f3
gguf : fix potential infinite for-loop (#4600)
texmex76 committed 1 year ago
Verified c30b1ef3
metal : refactor kernel loading code (#4794)
ggerganov committed 1 year ago
Verified b38b5e93
compare-llama-bench: tweak output format (#4910)
JohannesGaessler committed 1 year ago
Verified 7dc78764
server : fix deadlock that occurs in multi-prompt scenarios (#4905)
ziedbha committed 1 year ago
Verified 356327fe
server : fix crash with multimodal models without BOS token (#4904)
makomk committed 1 year ago
Verified ee8243ad
convert : update phi-2 to latest HF repo (#4903)
ggerganov committed 1 year ago
Verified 15ebe592
sync : ggml
ggerganov committed 1 year ago
Verified de473f5f
ggml : fix 32-bit ARM compat for IQ2_XS (whisper/1758)
ggerganov committed 1 year ago
Verified f2384612
backend_sched : fix assignments
slaren committed 1 year ago
Verified fa5c1fb4
examples : add pydantic models to GBNF grammar generator (#4883)
Maximilian-Winter committed 1 year ago
Verified 52ee4540
CUDA: faster q8_0 -> f16 dequantization (#4895)
JohannesGaessler committed 1 year ago
Verified 3fe81781
llama : ggml-backend integration (#4766)
slaren committed 1 year ago
Verified e7e4df03
llama : remove redundant assert for StableLM (#4901)
ggerganov committed 1 year ago
Verified 584d674b
export-lora : use LLAMA_FILE_MAGIC_GGLA (#4894)
danbev committed 1 year ago
Verified 930f907d
llama.swiftui : update models layout (#4826)
isaiahbjork committed 1 year ago
Verified e790eef2
gitignore : imatrix
ggerganov committed 1 year ago
Verified 5537d9d3
CUDA: fix softmax compile for old CUDA versions (#4862)
JohannesGaessler committed 1 year ago
Verified 1b280c9f
Terms Of Use Privacy Policy Refund Policy Impressum