 lhez
|
bee28421be
opencl : broadcast for soft_max (#14510)
|
il y a 6 mois |
 Jeff Bolz
|
2b72bedec1
vulkan: support mixed/deepseekR1 FA head sizes (#14509)
|
il y a 6 mois |
 Johannes Gäßler
|
c8c4495b8d
ggml: backward pass for split swiglu (#14483)
|
il y a 6 mois |
 Nicolò Scipione
|
7b63a71a6b
Fix conditional enabling following arch checks for ggml-sycl (#14504)
|
il y a 6 mois |
 Xuan-Son Nguyen
|
0c2ee38ab7
convert : correct gemma 3n conversion (#14450)
|
il y a 6 mois |
 Georgi Gerganov
|
a70c8a0c4b
kv-cache : use ggml_set_rows (#14285)
|
il y a 6 mois |
|
Georgi Gerganov
|
9067487c44
ggml : fix FA mask dim 2 and 3 (#14505)
|
il y a 6 mois |
|
Georgi Gerganov
|
d4cdd9c1c3
ggml : remove kompute backend (#14501)
|
il y a 6 mois |
 Aman Gupta
|
55c2646b45
CUDA: add dynamic shared mem to softmax, refactor general usage (#14497)
|
il y a 6 mois |
 Sigbjørn Skjæret
|
e75ba4c043
gguf-py : add support for chat template jinja files (#14508)
|
il y a 6 mois |
 compilade
|
5d46babdc2
llama : initial Mamba-2 support (#9126)
|
il y a 6 mois |
|
Georgi Gerganov
|
e17991c466
sync : ggml
|
il y a 6 mois |
 Daniel Bevenius
|
c46944aa25
ggml : add version function to get lib version (ggml/1286)
|
il y a 6 mois |
 Rotem Dan
|
f3ed38d793
Set RPATH to "@loader_path" / "$ORIGIN" to ensure executables and dynamic libraries search for dependencies in their origin directory. (#14309)
|
il y a 6 mois |
|
Aman Gupta
|
55a1c5a5fd
CUDA: add softmax broadcast (#14475)
|
il y a 6 mois |
|
Johannes Gäßler
|
12a81af45f
CUDA: broadcasting for FlashAttention mask (#14500)
|
il y a 6 mois |
|
Jeff Bolz
|
8875523eb3
vulkan: support softmax/FA batch and broadcast (#14449)
|
il y a 6 mois |
|
Georgi Gerganov
|
ec68e84c32
ggml : support bcast ggml_soft_max_ext, ggml_flash_attn_ext (#14435)
|
il y a 6 mois |
 zhouwg
|
307e79d33d
opencl : fix possible buffer overflow in dump_tensor (#14490)
|
il y a 6 mois |
|
Georgi Gerganov
|
d7f5f4e578
simple-chat : fix context-exceeded condition (#14494)
|
il y a 6 mois |
 Eric Zhang
|
c8a4e470f6
opencl : skip empty nodes on cgraph compute (#14491)
|
il y a 6 mois |
|
lhez
|
603e43dc91
opencl : update upscale to support align corners (#14488)
|
il y a 6 mois |
|
Sigbjørn Skjæret
|
611ba4b264
ci : add OpenCL to labeler workflow (#14496)
|
il y a 6 mois |
|
Eric Zhang
|
85841e121d
github : add OpenCL backend to issue templates (#14492)
|
il y a 6 mois |
 Björn Ganster
|
68b3cd6514
ggml : Callback before abort (#14481)
|
il y a 6 mois |
|
Georgi Gerganov
|
de56944147
ci : disable fast-math for Metal GHA CI (#14478)
|
il y a 6 mois |
 Grzegorz Grasza
|
1b2aaf28ac
Add Vulkan images to docker.md (#14472)
|
il y a 6 mois |
 Chenguang Li
|
343b6e94b6
CANN: update aclnnGroupedMatmulV2 to aclnnGroupedMatmulV3 (#14411)
|
il y a 6 mois |
|
Jeff Bolz
|
6a746cf9c4
vulkan: Split large mul_mat_id to fit in shared memory (#14451)
|
il y a 6 mois |
|
Sigbjørn Skjæret
|
eff5e45443
add GELU_ERF (#14455)
|
il y a 6 mois |
