Georgi Gerganov
|
4d0dcd4a06
cuda : fix rope with partial rotation and non-cont src (#14580)
|
преди 6 месеца |
Aman Gupta
|
75c91de6e9
CUDA: add bilinear interpolation for upscale (#14563)
|
преди 6 месеца |
R0CKSTAR
|
68155c66f0
musa: fix build warnings (unused variable) (#14561)
|
преди 6 месеца |
Sigbjørn Skjæret
|
e1a7059053
llama : fix incorrect minicpm3 v_states shape (#14571)
|
преди 6 месеца |
Sigbjørn Skjæret
|
12f55c302b
llama : remove ggml_cont where possible (#14568)
|
преди 6 месеца |
Aman Gupta
|
b9c3eefde1
CUDA: add bf16 and i32 to getrows (#14529)
|
преди 6 месеца |
Eve
|
6491d6e4f1
vulkan: increase LOAD_VEC_A to 8 (IQ1/IQ2) or 4 (IQ3) (#14485)
|
преди 6 месеца |
Jeff Bolz
|
e592be1575
vulkan: fix rms_norm+mul fusion (#14545)
|
преди 6 месеца |
Jeff Bolz
|
a0374a67e2
vulkan: Handle updated FA dim2/3 definition (#14518)
|
преди 6 месеца |
Sigbjørn Skjæret
|
ddef99522d
server : fix assistant prefilling when content is an array (#14360)
|
преди 6 месеца |
Sigbjørn Skjæret
|
6681688146
opencl: add GELU_ERF (#14476)
|
преди 6 месеца |
Georgi Gerganov
|
bac8bed248
eval-callback : check for empty input (#14539)
|
преди 6 месеца |
R0CKSTAR
|
b81510a7b7
test-backend-ops: add support for specifying output format (#14368)
|
преди 6 месеца |
Georgi Gerganov
|
ef797db357
metal : disable fast math in all quantize kernels (#14528)
|
преди 6 месеца |
Georgi Gerganov
|
67d1ef23c6
batch : add optional for sequential equal split (#14511)
|
преди 6 месеца |
Georgi Gerganov
|
7b50f7c025
graph : prepare for 4D mask (#14515)
|
преди 6 месеца |
Georgi Gerganov
|
c79184d2d1
batch : add n_used count (#14512)
|
преди 6 месеца |
luyhcsu
|
499a8f5a78
CANN: Replace aclrtMemsetSync with aclnnInplaceZero operator (#14002)
|
преди 6 месеца |
Sigbjørn Skjæret
|
28657a8229
ggml : implement GEGLU_ERF and GEGLU_QUICK ops (#14445)
|
преди 6 месеца |
lhez
|
bee28421be
opencl : broadcast for soft_max (#14510)
|
преди 6 месеца |
Jeff Bolz
|
2b72bedec1
vulkan: support mixed/deepseekR1 FA head sizes (#14509)
|
преди 6 месеца |
Johannes Gäßler
|
c8c4495b8d
ggml: backward pass for split swiglu (#14483)
|
преди 6 месеца |
Nicolò Scipione
|
7b63a71a6b
Fix conditional enabling following arch checks for ggml-sycl (#14504)
|
преди 6 месеца |
Xuan-Son Nguyen
|
0c2ee38ab7
convert : correct gemma 3n conversion (#14450)
|
преди 6 месеца |
Georgi Gerganov
|
a70c8a0c4b
kv-cache : use ggml_set_rows (#14285)
|
преди 6 месеца |
Georgi Gerganov
|
9067487c44
ggml : fix FA mask dim 2 and 3 (#14505)
|
преди 6 месеца |
Georgi Gerganov
|
d4cdd9c1c3
ggml : remove kompute backend (#14501)
|
преди 6 месеца |
Aman Gupta
|
55c2646b45
CUDA: add dynamic shared mem to softmax, refactor general usage (#14497)
|
преди 6 месеца |
Sigbjørn Skjæret
|
e75ba4c043
gguf-py : add support for chat template jinja files (#14508)
|
преди 6 месеца |
compilade
|
5d46babdc2
llama : initial Mamba-2 support (#9126)
|
преди 6 месеца |