Jeff Bolz
|
80dd7ff22f
vulkan: Optimize contiguous copies (#10254)
|
пре 1 година |
Georgi Gerganov
|
841f27abdb
metal : optimize FA kernels (#10171)
|
пре 1 година |
Zhiyuan Li
|
3bcd40b3c5
Optimize RWKV6 Operator Naming and Implement Multi-core CPU/ SYCL Acceleration (#10133)
|
пре 1 година |
Georgi Gerganov
|
5c333e0140
metal : add BF16 support (#8439)
|
пре 1 година |
Diego Devesa
|
9f40989351
ggml : move CPU backend to a separate file (#10144)
|
пре 1 година |
Johannes Gäßler
|
c39665f589
CUDA: fix MMQ for non-contiguous src0, add tests (#10021)
|
пре 1 година |
Johannes Gäßler
|
80273a306d
CUDA: fix 1D im2col, add tests (ggml/993)
|
пре 1 година |
Jun Hee Yoo
|
4c9388fb96
metal : add POOL2D and fix IM2COL (#9943)
|
пре 1 година |
Diego Devesa
|
dca1d4b58a
ggml : fix BLAS with unsupported types (#9775)
|
пре 1 година |
Diego Devesa
|
6374743747
ggml : add backend registry / device interfaces to BLAS backend (#9752)
|
пре 1 година |
Johannes Gäßler
|
fabdc3bda3
ggml/ex: calculate accuracy in graph, adapt MNIST (ggml/980)
|
пре 1 година |
Diego Devesa
|
c83ad6d01e
ggml-backend : add device and backend reg interfaces (#9707)
|
пре 1 година |
Johannes Gäßler
|
e98c1c188e
test: fix OPT_STEP_ADAMW for test-backend-ops (ggml/974)
|
пре 1 година |
Johannes Gäßler
|
7254cdf7e8
ggml: fix gradient allocation logic (ggml/966)
|
пре 1 година |
slaren
|
1b2f992cd2
test-backend-ops : use flops for some performance tests (#9657)
|
пре 1 година |
Johannes Gäßler
|
a5b57b08ce
CUDA: enable Gemma FA for HIP/Pascal (#9581)
|
пре 1 година |
Molly Sophia
|
2a63caaa69
RWKV v6: RWKV_WKV op CUDA implementation (#9454)
|
пре 1 година |
Johannes Gäßler
|
424c5d00a9
ggml/examples: add backend support for numerical optimization (ggml/949)
|
пре 1 година |
Georgi Gerganov
|
d6a04f872d
ggml : hide ggml_object, ggml_cgraph, ggml_hash_set (#9408)
|
пре 1 година |
Georgi Gerganov
|
a876861455
metal : update support condition for im2col + fix warning (#0)
|
пре 1 година |
Johannes Gäßler
|
202084d31d
tests: add gradient tests for all backends (ggml/932)
|
пре 1 година |
Salvatore Mesoraca
|
efe6a83e30
ggml : fix cont with transposed tensors when one dimension is 1 (ggml/934)
|
пре 1 година |
compilade
|
9bc6db28d0
ggml-quants : ternary packing for TriLMs and BitNet b1.58 (#8151)
|
пре 1 година |
Georgi Gerganov
|
231cff5f6f
sync : ggml
|
пре 1 година |
Georgi Gerganov
|
fc18425b6a
ggml : add SSM Metal kernels (#8546)
|
пре 1 година |
slaren
|
0c41e03ceb
metal : gemma2 flash attention support (#9159)
|
пре 1 година |
Johannes Gäßler
|
e11bd856d5
CPU/CUDA: Gemma 2 FlashAttention support (#8542)
|
пре 1 година |
zhentaoyu
|
4f8d19ff17
[SYCL] Fix SYCL `im2col` and `convert` Overflow with Large Dims (#9052)
|
пре 1 година |
Molly Sophia
|
2d5dd7bb3f
ggml : add epsilon as a parameter for group_norm (#8818)
|
пре 1 година |
0cc4m
|
064cdc265f
vulkan : fix Qantized Mat-Vec Mul on AMD GPUs for ncols < 64 (#8855)
|
пре 1 година |