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

b4958

ef19c717 · run: de-duplicate fmt and format functions and optimize (#11596) · Mar 25, 2025
b4957

053b3f9a · ggml-cpu : update KleidiAI to v1.5.0 (#12568) · Mar 25, 2025
b4956

e2f56017 · SYCL: disable Q4_0 reorder optimization (#12560) · Mar 25, 2025
b4953

2d77d88e · context : fix worst-case reserve outputs (#12545) · Mar 25, 2025
b4951

2b65ae30 · opencl: simplify kernel embedding logic in cmakefile (#12503) · Mar 24, 2025
b4948

00d53800 · llama-vocab : add SuperBPE pre-tokenizer (#12532) · Mar 24, 2025
b4947

7ea75035 · CUDA: Fix clang warnings (#12540) · Mar 24, 2025
b4946

c54f6b79 · mmap : skip resource limit checks on AIX (#12541) · Mar 24, 2025
b4945

9b169a4d · vulkan: fix mul_mat_vec failure in backend tests (#12529) · Mar 24, 2025
b4944

77f9c6bb · server : Add verbose output to OAI compatible chat endpoint. (#12246) · Mar 23, 2025
b4942

fbdfefe7 · llama : gemma3 : use output tensor if it exists in model weight (#12506) · Mar 22, 2025
b4940

fac63a3d · musa: refine compute capability (#12493) · Mar 22, 2025
b4939

eddfb438 · vulkan: Optimize mul_mat_vec p021 and nc shaders (#12505) · Mar 22, 2025
b4938

4375415b · Vulkan: RTE rounding for cpy to quant (#12480) · Mar 21, 2025
b4937

30c42ef5 · vulkan: workaround for AMD Windows driver 16 bit unpack8 bug (#12472) · Mar 21, 2025
b4936

af04481e · model : do not repack if a GPU device is present (#12498) · Mar 21, 2025
b4935

960e7260 · chore : cleanup llama_model_loader::TENSOR_ usage (#12492) · Mar 21, 2025
b4934

ea1518e8 · llama-tts : avoid crashes related to bad model file paths (#12482) · Mar 21, 2025
b4933

1aa87ee5 · [SYCL] Fix build on Windows when ccache enabled (#9954) (#9976) · Mar 21, 2025
b4932

9ffcc9e3 · sycl: cleanup oneDNN related code (#12097) · Mar 21, 2025

1
…
12
13
14
15
16
17
18
19
20
…
178