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

b1671

8fe03ffd · common : remove incorrect --model-draft default (#4568) · Dec 21, 2023
b1670

91544948 · CUDA: mul_mat_id always on GPU for batches >= 32 (#4553) · Dec 21, 2023
b1667

66f35a2f · cuda : better error message for ggml_get_rows (#4561) · Dec 21, 2023
b1666

13988239 · cuda : replace asserts in wrong architecture checks with __trap (#4556) · Dec 21, 2023
b1665

d3223afd · llama : disable per-tensor info prints on model load (#4562) · Dec 21, 2023
b1664

1d7a1912 · Fix access violation in ggml_cuda_free_data if tensor->extra is NULL (#4554) · Dec 21, 2023
b1663

799fc226 · CUDA: Faster Mixtral prompt processing (#4538) · Dec 20, 2023
b1662

328b83de · ggml : fixed check for _MSC_VER (#4535) · Dec 19, 2023
b1661

a7aee47b · ggml-cuda: Fix HIP build (#4528) · Dec 18, 2023
b1660

0e18b2e7 · llama.swiftui : add tinyllama 1.1B F16 · Dec 18, 2023
b1659

6ff39b12 · llama.swiftui : add more models · Dec 18, 2023
b1658

b9e74f9b · llama : add phi-2 + fix NeoX rope + ggml_mul_mat_set_prec (#4490) · Dec 18, 2023
b1657

3c04bf6d · llama : fix try_override for bool_value which always return true (#4519) · Dec 18, 2023
b1656

2994f0c5 · decode : fix logits_valid for legacy API (#4516) · Dec 17, 2023
b1654

800a489e · llama.swiftui : add bench functionality (#4483) · Dec 17, 2023
b1652

919c4066 · build : Check the ROCm installation location (#4485) · Dec 17, 2023
b1651

45668633 · finetune : keep allocs alive until all allocations are done (#4486) · Dec 17, 2023
b1650

0ffc92d2 · server : disable llm logs if SERVER_VERBOSE is off (#3792) · Dec 17, 2023
b1649

8edd2b40 · server : fix grammar being ignored (#4494) · Dec 17, 2023
b1648

eb16dae7 · server : fix possible ambiguity in content type charset (#4501) · Dec 17, 2023