Skip to content
GitLab
Explore
Sign in
Tags
Tags give the ability to mark specific points in history as being important
b3033
2ab97728
·
sync : ggml
·
May 29, 2024
b3030
504f0c34
·
ggml : fix typo in ggml.c (#7603)
·
May 29, 2024
b3029
b864b50c
·
[SYCL] Align GEMM dispatch (#7566)
·
May 29, 2024
b3028
02c1ecad
·
Tokenizer WPM fixes (#7500)
·
May 28, 2024
b3027
6bd12ce4
·
sycl : fix assert (#7563)
·
May 28, 2024
b3026
5442939f
·
llama : support small Granite models (#7481)
·
May 28, 2024
b3025
56411a95
·
vulkan: properly initialize vulkan devices for LLAMA_SPLIT_MODE_NONE (#7552)
·
May 28, 2024
b3024
2b737caa
·
rpc : resource management rework (#7562)
·
May 28, 2024
b3023
ee3dff6b
·
Add support for DeepseekV2ForCausalLM (#7519)
·
May 28, 2024
b3021
8b99e2aa
·
llama : handle unknown utf8 bytes (#7588)
·
May 28, 2024
b3019
e2b06507
·
[SYCL]fix ggml_sycl_mul_mat_id() to match the change of api (#7436)
·
May 28, 2024
b3018
0548a418
·
ggml : generalize GGML_OP_CONCAT (#7563)
·
May 28, 2024
b3015
74b239b3
·
llava : update clip.h (#7580)
·
May 28, 2024
b3014
852aafb1
·
update HIP_UMA #7399 (#7414)
·
May 28, 2024
b3012
10b1e458
·
make: add --device-debug to NVCC debug flags (#7542)
·
May 27, 2024
b3011
197c0068
·
Allow multiple copy function pointers for CUDA graph kernel param updates (#7565)
·
May 27, 2024
b3010
95f84d5c
·
Fix q_xxs using mul_mat_q (#7459)
·
May 27, 2024
b3008
1d8fca72
·
metal : add GGML_OP_REPEAT kernels (#7557)
·
May 27, 2024
b3007
62bfef51
·
metal : disable FA kernel for HS=256 (#7556)
·
May 27, 2024
b3006
eaf6e031
·
llama : add comments about experimental flags (#7544)
·
May 27, 2024
Prev
1
2
3
4
5
6
7
…
99
Next