为了安全，强烈建议开启2FA双因子认证：User Settings -> Account -> Enable two-factor authentication！！！

Tags

Tags give the ability to mark specific points in history as being important

b3991

fc83a9e5 · ggml : add Q4_0_8_8 RISC-V GEMV and GEMM kernels (#10029) · Oct 30, 2024
b3990

c5b0f4b5 · llama : refactor model loader with backend registry (#10026) · Oct 30, 2024
b3989

8f275a7c · ggml: Add POOL2D OP for GPU acceleration to the Vulkan backend in the MobileVLM model. (#9763) · Oct 29, 2024
b3988

8d8ff715 · llama : remove Tail-Free sampling (#10071) · Oct 29, 2024
b3987

61715d5c · llama : Add IBM granite template (#10013) · Oct 28, 2024
b3985

524afeec · musa: workaround for Guilty Lockup in cleaning src0 (#10042) · Oct 28, 2024
b3984

8125e6cb · server : don't overfill the batch during infill (#10018) · Oct 28, 2024
b3983

8841ce3f · llama : switch KQ multiplication to F32 precision by default (#10015) · Oct 27, 2024
b3982

cc2983d3 · sync : ggml · Oct 26, 2024
b3978

ff252ea4 · llama : add DRY sampler (#9702) · Oct 25, 2024
b3977

d80fb71f · llama: string_split fix (#10022) · Oct 25, 2024
b3976

2f8bd2b9 · llamafile : extend sgemm.cpp support for Q5_0 models (#10010) · Oct 25, 2024
b3975

bc5ba007 · server : check that the prompt fits in the slot's context (#10030) · Oct 25, 2024
b3974

958367bf · server : refactor slot input data, move tokenizer to HTTP thread (#10023) · Oct 24, 2024
b3972

167a5156 · CUDA: fix insufficient buffer clearing for MMQ (#10032) · Oct 24, 2024
b3971

c39665f5 · CUDA: fix MMQ for non-contiguous src0, add tests (#10021) · Oct 24, 2024
b3970

0a1c750c · server : samplers accept the prompt correctly (#10019) · Oct 23, 2024
b3969

190a37d7 · sync : ggml · Oct 23, 2024
b3967

80273a30 · CUDA: fix 1D im2col, add tests (ggml/993) · Oct 23, 2024
b3965

ac113a0f · llama.vim : add classic vim support (#9995) · Oct 23, 2024

1
…
44
45
46
47
48
49
50
51
52
…
178