Admin message
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
b3644
42c76d13
·
Threadpool: take 2 (#8672)
·
Aug 30, 2024
b3643
9f7d4bcf
·
server : fix crash when error handler dumps invalid utf-8 json (#9195)
·
Aug 30, 2024
b3639
20f1789d
·
vulkan : fix build (#0)
·
Aug 27, 2024
b3636
78eb487b
·
llama : fix qs.n_attention_wv for DeepSeek-V2 (#9156)
·
Aug 27, 2024
b3635
a77feb5d
·
server : add some missing env variables (#9116)
·
Aug 27, 2024
b3634
2e59d61c
·
llama : fix ChatGLM4 wrong shape (#9194)
·
Aug 27, 2024
b3633
75e1dbba
·
llama : fix llama3.1 rope_freqs not respecting custom head_dim (#9141)
·
Aug 27, 2024
b3632
ad76569f
·
common : Update stb_image.h to latest version (#9161)
·
Aug 27, 2024
b3631
7d787ed9
·
ggml : do not crash when quantizing q4_x_x with an imatrix (#9192)
·
Aug 26, 2024
b3630
06658ad7
·
metal : separate scale and mask from QKT in FA kernel (#9189)
·
Aug 26, 2024
b3629
fc18425b
·
ggml : add SSM Metal kernels (#8546)
·
Aug 26, 2024
b3628
879275ac
·
tests : fix compile warnings for unreachable code (#9185)
·
Aug 26, 2024
b3627
7a3df798
·
ci : add VULKAN support to ggml-ci (#9055)
·
Aug 26, 2024
b3625
0c41e03c
·
metal : gemma2 flash attention support (#9159)
·
Aug 26, 2024
b3623
436787f1
·
llama : fix time complexity of string replacement (#9163)
·
Aug 26, 2024
b3622
93bc3839
·
common: fixed not working find argument --n-gpu-layers-draft (#9175)
·
Aug 26, 2024
b3621
f91fc563
·
CUDA: fix Gemma 2 numerical issues for FA (#9166)
·
Aug 25, 2024
b3620
e11bd856
·
CPU/CUDA: Gemma 2 FlashAttention support (#8542)
·
Aug 24, 2024
b3619
8f824ffe
·
quantize : fix typo in usage help of `quantize.cpp` (#9145)
·
Aug 24, 2024
b3618
3ba780e2
·
lora : fix llama conversion script with ROPE_FREQS (#9117)
·
Aug 23, 2024
1
…
56
57
58
59
60
61
62
63
64
…
178