Skip to content
GitLab
Explore
Sign in
Tags
Tags give the ability to mark specific points in history as being important
b2916
b43272af
·
Unicode codepoint flags for custom regexs (#7245)
·
May 18, 2024
b2915
0fc1e820
·
CUDA: faster large batch FA without tensor cores (#7314)
·
May 17, 2024
b2914
82ca83db
·
ROCm: use native CMake HIP support (#5966)
·
May 17, 2024
b2913
f4bd8b3d
·
rpc : set SO_REUSEADDR for the server socket (#7320)
·
May 17, 2024
b2910
27b04069
·
llama : use n_embd_head_v when reshaping kqv (#7327)
·
May 17, 2024
b2909
29c60d8c
·
tokenization: add warning for double BOS (#7332)
·
May 17, 2024
b2908
359cbe3f
·
ggml-quants, llama : removed excess checks (#7274)
·
May 17, 2024
b2906
ee94172d
·
server : add support for the RPC backend (#7305)
·
May 17, 2024
b2901
3b3963c5
·
rpc : add command line arg for specifying backend memory
·
May 16, 2024
b2899
0350f581
·
grammar, json, llama: replace push on emplace if it possible (#7273)
·
May 16, 2024
b2897
172b7821
·
ci: fix bin/Release path for windows-arm64 builds (#7317)
·
May 16, 2024
b2894
e1b40ac3
·
ggml : use dynamic thread scheduling for matrix multiplication (#6915)
·
May 15, 2024
b2893
dc020985
·
Avoid unnecessarily disabling CUDA graphs (#7302)
·
May 15, 2024
b2892
344f9126
·
ggml : tag ggml_tensor::backend as deprecated (#7290)
·
May 15, 2024
b2891
9a17ab91
·
Add missing " (#7303)
·
May 15, 2024
b2890
ea3b0590
·
embedding : free the batch after execution (#7297)
·
May 15, 2024
b2889
29499bb5
·
sync : ggml
·
May 15, 2024
b2885
e8a7fd4f
·
metal : support FA without mask + add asserts (#7278)
·
May 14, 2024
b2884
a5e3fde8
·
sync : ggml
·
May 14, 2024
b2879
4f026363
·
server: free sampling contexts on exit (#7264)
·
May 14, 2024
Prev
1
2
3
4
5
6
7
8
9
10
…
98
Next