Commits

Commits on May 24, 2025

Merge remote-tracking branch 'origin/master' into tool-diffs
ochafik
committed
40951c8
Copy full SHA for 40951c8

Commits on May 23, 2025

ci : enable winget package updates (#13734)
slaren
authored
b775345
Copy full SHA for b775345
ci : add winget package updater (#13732)
slaren
authored
a70a8a6
Copy full SHA for a70a8a6
hparams : initialize arrays (#13728)
Show description for d13d0f6
ggerganov
authored
d13d0f6
Copy full SHA for d13d0f6
llama : allow custom list of swa_layers (#13726)
ngxson
authored
8a2afb7
Copy full SHA for 8a2afb7
Merge remote-tracking branch 'origin/master' into tool-diffs
ochafik
committed
f0d5df2
Copy full SHA for f0d5df2
server : support audio input (#13714)
Show description for 9ecf3e6
ngxson
authored
9ecf3e6
Copy full SHA for 9ecf3e6
CANN: Support MUL_MAT_ID for q8_0 and q4_0 (#13705)
Show description for faaaff5
noemotiovon
authored
faaaff5
Copy full SHA for faaaff5
ggml : fix the order of ggml_unary_op (#13718)
ngxson
authored
e16c473
Copy full SHA for e16c473
vulkan: support CPY from any type to itself (#13695)
Show description for 1dcd019
jeffbolznv
authored
1dcd019
Copy full SHA for 1dcd019
vulkan: Disable coopmat/coopmat2/bfloat extensions if glslc doesn't support it (#13696)
jeffbolznv
authored
c10ed6c
Copy full SHA for c10ed6c
use LOG_WARN to replace `std::cerr` (#13657)
foldl
authored
a127ff1
Copy full SHA for a127ff1

Commits on May 22, 2025

release : fix windows hip release (#13707)
Show description for 3079e9a
slaren
authored
3079e9a
Copy full SHA for 3079e9a
tts : fix n_ubatch + make WavTokenizer cache-less (#13713)
Show description for 8a1d206
ggerganov
authored
8a1d206
Copy full SHA for 8a1d206
mtmd : add ultravox audio input (#13623)
Show description for 797990c
ngxson
authored
797990c
Copy full SHA for 797990c
common: Include torch package for s390x (#13699)
Show description for ab86335
taronaeo
authored
ab86335
Copy full SHA for ab86335
server : pad small embedding batches (#13692)
Show description for cc74d5b
ggerganov
authored
cc74d5b
Copy full SHA for cc74d5b
gguf-py : correct charsmap parameter typing (#13701)
CISC
authored
5be24af
Copy full SHA for 5be24af
sycl : Remove waits from function calls (#13702)
Show description for d394a9a
s-Nick
authored
d394a9a
Copy full SHA for d394a9a
SYCL: Avoid using with SYCL-Graph for unsupported nodes (#13587)
Show description for 6b56a64
EwanC
authored
6b56a64
Copy full SHA for 6b56a64

Commits on May 21, 2025

opencl: Add support for multiple devices (#12622)
Show description for a4e8912
linehill
authored
a4e8912
Copy full SHA for a4e8912
opencl: fix couple crashes (#12795)
Show description for edbf42e
linehill
authored
edbf42e
Copy full SHA for edbf42e
releases : build CPU backend separately (windows) (#13642)
slaren
authored
d643bb2
Copy full SHA for d643bb2
hparams : support models for which all layers use SWA (#13682)
Show description for 8e186ef
ggerganov
authored
8e186ef
Copy full SHA for 8e186ef
server : improve error reporting (#13680)
ggerganov
authored
5fbfe38
Copy full SHA for 5fbfe38
convert : add qwen2vl support for unsloth merges (#13686)
antichristHater
authored
c76532e
Copy full SHA for c76532e
examples : switch retrieval to llama_encode (#13685)
Show description for 2aa777d
CISC
authored
2aa777d
Copy full SHA for 2aa777d
gguf-py : display the invalid gguf type (#13687)
Show description for eb0f5c2
emmanuel-ferdman
authored
eb0f5c2
Copy full SHA for eb0f5c2
ggml : add ggml_gelu_erf() (#13667)
Show description for cf4cb59
ngxson
authored
cf4cb59
Copy full SHA for cf4cb59
server : Add the endpoints /api/tags and /api/chat (#13659)
Show description for 0d5c742
R-Dson
authored
0d5c742
Copy full SHA for 0d5c742
server : fix first message identification (#13634)
Show description for 42158ae

doringeman
and
p1-0tr
authored
42158ae
Copy full SHA for 42158ae
kv-cache : simplify the interface (#13660)
Show description for 797f2ac
ggerganov
authored
797f2ac
Copy full SHA for 797f2ac
model : disable SWA for Phi models (#13676)
Show description for b44890d
ggerganov
authored
b44890d
Copy full SHA for b44890d
musa: Upgrade MUSA SDK version to rc4.0.1 and use mudnn::Unary::IDENTITY op to accelerate D2D memory copy (#13647)
Show description for 3398305

yeahdongcn
and
JohannesGaessler
authored
3398305
Copy full SHA for 3398305

Commits on May 20, 2025

vulkan: fix warnings (#13626)
Show description for fb1cab2
netrunnereve
authored
fb1cab2
Copy full SHA for fb1cab2