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

b4281

c2a16c0b · server : fix free of spec context and batch (#10651) · Dec 07, 2024
b4280

3df784b3 · Vulkan: VK_KHR_cooperative_matrix support to speed up prompt processing (#10597) · Dec 07, 2024
b4279

86a19349 · metal : Extend how Llama.cpp locates metal resources (#10676) · Dec 07, 2024
b4276

f162d45a · common : bring back --no-warmup to server (#10686) · Dec 06, 2024
b4273

c9c6e01d · vulkan: Add VK_NV_cooperative_matrix2 support for mul_mat and flash attention (#10206) · Dec 05, 2024
b4272

6fe62478 · llama : add Minerva 7B model support (#10673) · Dec 05, 2024
b4271

0cd182eb · sync : ggml · Dec 05, 2024
b4267

f112d198 · Update deprecation-warning.cpp (#10619) · Dec 04, 2024
b4266

1da7b765 · server : fix speculative decoding with context shift (#10641) · Dec 04, 2024
b4265

59f4db10 · ggml : add predefined list of CPU backend variants to build (#10626) · Dec 04, 2024
b4262

8d0cfd55 · llama: Support MiniCPM-1B (with & w/o longrope) (#10559) · Dec 04, 2024
b4261

2759916d · vulkan: Implement "fast divide" (mul+shift) for unary ops like copy (#10642) · Dec 04, 2024
b4260

40c6d79f · SYCL : Move to compile time oneMKL interface backend selection for NVIDIA backend (#10584) · Dec 04, 2024
b4258

cd2f37b3 · Avoid using __fp16 on ARM with old nvcc (#10616) · Dec 04, 2024
b4256

01e6d9bb · clip : add sycl support (#10574) · Dec 04, 2024
b4255

cc98896d · vulkan: optimize and reenable split_k (#10637) · Dec 03, 2024
b4254

91c36c26 · server : (web ui) Various improvements, now use vite as bundler (#10599) · Dec 03, 2024
b4253

1cd3df46 · scripts : remove amx sync · Dec 03, 2024
b4248

3b4f2e33 · llama : add missing LLAMA_API for llama_chat_builtin_templates (#10636) · Dec 03, 2024
b4246

0115df2f · metal : small-batch mat-mul kernels (#10581) · Dec 03, 2024

1
…
35
36
37
38
39
40
41
42
43
…
178