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

b3808

1e7b9299 · ggml : AVX512 gemm for Q4_0_8_8 (#9532) · Sep 23, 2024
b3807

37f8c7b4 · perplexity : remove extra new lines after chunks (#9596) · Sep 23, 2024
b3806

bf9c1013 · metal : use F32 prec for K*Q in vec FA (#9595) · Sep 23, 2024
b3805

e62e9789 · Revert "[SYCL] fallback mmvq (#9088)" (#9579) · Sep 23, 2024
b3804

c35e586e · musa: enable building fat binaries, enable unified memory, and disable Flash... · Sep 22, 2024
b3803

912c331d · Fix merge error in #9454 (#9589) · Sep 22, 2024
b3802

a5b57b08 · CUDA: enable Gemma FA for HIP/Pascal (#9581) · Sep 22, 2024
b3801

ecd5d6b6 · llama: remove redundant loop when constructing ubatch (#9574) · Sep 22, 2024
b3800

2a63caaa · RWKV v6: RWKV_WKV op CUDA implementation (#9454) · Sep 22, 2024
b3799

d09770ca · ggml-alloc : fix list of allocated tensors with GGML_ALLOCATOR_DEBUG (#9573) · Sep 21, 2024
b3798

41f47787 · Update CUDA graph on scale change plus clear nodes/params (#9550) · Sep 21, 2024
b3797

e948a7da · CI: Provide prebuilt windows binary for hip (#9467) · Sep 21, 2024
b3796

63351143 · quantize : improve type name parsing (#9570) · Sep 20, 2024
b3795

d13edb17 · ggml : fix builds (#0) · Sep 20, 2024
b3790

5cb12f68 · CUDA: fix sum.cu compilation for CUDA < 11.7 (#9562) · Sep 20, 2024
b3789

d39e2674 · examples : flush log upon ctrl+c (#9559) · Sep 20, 2024
b3788

722ec1eb · perplexity : do not escape input data by default (#9548) · Sep 20, 2024
b3787

6026da52 · server : clean-up completed tasks from waiting list (#9531) · Sep 19, 2024
b3786

eca0fab4 · imatrix : disable prompt escape by default (#9543) · Sep 19, 2024
b3785

64c6af31 · ggml : fix n_threads_cur initialization with one thread (#9538) · Sep 18, 2024

1
…
50
51
52
53
54
55
56
57
58
…
178