HIP: implement FlashAttention via rocWMMA for CDNA and RDNA3+ (llama/12032)
a027c1d
David Huang
commited on
Spaces:
Running
Running
Commit History
ggml : fix kleidiai build (llama/12159)
dbc0180
ag2s20150909
commited on
SYCL: Move CPY kernels to a separate file and add few missing kernels (llama/12133)
1d6d451
Akarshan Biswas
commited on
ggml-backend : keep paths in native string type when possible (llama/12144)
6e89d8c
Diego Devesa
commited on
CUDA: compress mode option and default to size (llama/12029)
4ec988a
Erik Scholz
commited on
ggml : upgrade init_tensor API to return a ggml_status (llama/11854)
d6b6852
William Tambellini
slaren
commited on
vulkan: add specific MMV kernels for IQ2 and IQ3 quants + optimizations (llama/11595)
d7d82b9
Rémy O
commited on
CUDA: fix logic for V100 + GGML_CUDA_FORCE_MMQ (llama/12098)
0b52fcc
ggml: aarch64: implement SVE kernels for q2_k_q8_k vector dot (llama/12064)
459beb1
Prashant Vithule
vithulep
commited on
CANN: Fix build error with GCC 13 (llama/11990)
dcf68db
vulkan: matmul dequantization improvements (llama/12015)
ffdf466
Eve
commited on
vulkan: improve im2col (llama/11826)
f6cff0a
Daniele
commited on
cmake: Fix ggml backend dependencies and installation (llama/11818)
c6c2a2c
Vladimir Vuksanovic
commited on
vulkan: fix assertion when qy_needs_dequant (llama/12068)
271c7e4
ggml-cpu: Fix build with sve (llama/12059)
4be146e
cuda: unary ops as float + de-duplicate (ggml/1130)
4bec2e4
cmdr2
commited on
cuda/vulkan: specify fp32-only support for some operations in supports_op (ggml/1129)
f959b90
cmdr2
commited on
cuda/cpu: Increase support for fp16 unary operations (ggml/1125)
67e8c32
cmdr2
commited on
Told cmake to install ggml-cpp.h as a public header file. (ggml/1126)
3d4f29c
petterreinholdtsen
Petter Reinholdtsen
commited on
common : more general m_audio_len update logic (#2855)
4674264
unverified
Ivy233
Ivy233
commited on
go : improve model download (#2756)
168712d
unverified
Ryan Johnson
commited on
common : fix audio loading by miniaudio (#2862)
494fb84
unverified
Dmitry Atamanov
commited on
fix: missing include common-whisper (#2858)
2271d56
unverified
Lin Xiaodong
commited on
ruby : follow audio library change (#2851)
b94e7d3
unverified
whisper : support GGML_BACKEND_DL (#2843)
2e6437e
unverified
common : separate whisper sources (#2846)
0447b9d
unverified
common : fix build min/max (#2845)
07533a2
unverified
examples : use miniaudio for direct decoding flac, mp3, ogg and wav (#2759)
7a280a4
unverified
Dmitry Atamanov
commited on
stream : stop on ^C when no audio is received (#2822)
45399ad
unverified
petterreinholdtsen
Petter Reinholdtsen
commited on
sync : ggml
7926873
Support pure float16 add/sub/mul/div operations in the CUDA (and CPU) backend (ggml/1121)
2b94a24
cmdr2
commited on
opencl: fix for small models (llama/11950)
4532dc6
lhez
Shawn Gu
Skyler Szot
commited on
Optimize mul_mat for Q4_0 on Intel GPU (llama/12035)
14fd317
Neo Zhang Jianyu
arthw
commited on
SYCL: Fix GGML_SYCL_DEBUG macro (llama/11995)
310a36c
ggml-cpu: Support s390x SIMD Instruction Set (llama/12019)
4aa54ec
Aaron Teo
Jinyang He
junchao-zhao
commited on
CUDA: app option to compile without FlashAttention (llama/12025)
fbc5f16
CUDA: optimize FA for GQA + large batches (llama/12014)
6662d54
cuda: Add Q5_1, Q5_0, Q4_1 and Q4_0 to F32 conversion support. (llama/12000)
6cb8158
CUDA: correct the lowest Maxwell supported by CUDA 12 (llama/11984)
6641178
MUSA: support ARM64 and enable dp4a .etc (llama/11843)
ab96dac
Bodhi
Bodhi Hu
commited on
ggml-cpu: Add CPU backend support for KleidiAI library (llama/11390)
9de6d81
Charles Xu
commited on
ggml: aarch64: implement SVE kernels for q3_K_q8_K vector dot (llama/11917)
1a1acd2
CUDA: use async data loading for FlashAttention (llama/11894)
5b9980d
vulkan: implement several ops relevant for ggml_opt (llama/11769)
3c2171d
Rémy O
commited on
vulkan: support multi/vision rope, and noncontiguous rope (llama/11902)
1c7a669
metal : fix the crash caused by the lack of residency set support on Intel Macs. (llama/11904)
afbd891
Hale Chan
commited on
metal : optimize dequant q6_K kernel (llama/11892)
376cbe6
Adrian Kretz
commited on
repo : update links to new url (llama/11886)
9705bb5
vulkan: initial support for IQ1_S and IQ1_M quantizations (llama/11528)
0d2e888
Rémy O
commited on