Morph MORPH
SIGN IN SIGN UP
dhruvhead / Paddle UNCLAIMED
0 0 1 C++

COMMITS

develop
August 24, 2026
苍
Optimize parallel third-party submodule initialization (#79679)
苍天荒 committed
2a52cd6
L
Fix the double registration of the shared layer in VPP mode (#79678)
Liumengyuan committed
4306edd
Y
Atten res cache (#79517)
yong yao committed
1f6018c
F
[Operator Mechanism] Add int64-safe dimension handling for GEMM and GEMV (#79601)
feixi committed
fb632d7
A
[Distributed] Support NCCL symmetric memory windows and CTAPolicy=ZERO (#79677)
adam-xiaoyao committed
b8eef5f
August 20, 2026
苍
[Operator Mechanism] fix zero-size tensor NaNs in kernels (#79586)
苍天荒 committed
3be2352
O
[Operator Mechanism]Check the guard in IndexType dispatch of big tensors (#79645)
omoYang committed
63b1f45
O
[Operator Mechanism] Fix int32 index overflow in pooling GPU kernels (#79608)
omoYang committed
811fbbe
G
[PHI][Build] isolate oneDNN headers from CUDA translation units (#79659)
gouzil committed
cca182d
Z
[FSDP] Support FSDP with EP (#79640)
ZhenxingLi committed
3cbc0a8
August 19, 2026
S
[Performance] paddle.save supports protocol=5 with zero-copy (#79660)
Shuhao Liang committed
2fd3134
R
[Build] Support CUDA 13.2 on Windows (#79603)
Runming Xie committed
17780a6
August 18, 2026
G
[CI] Fix git clone cache invalidation (#79671)
Gu Shiwei committed
ce1f229
L
[Execute Infrastructure] Add Paddle AI review skill (#79669)
liuhao2638 committed
a993ad4
August 17, 2026
N
[CI] Remove checkout from skill path checks (#79666)
Nyako Shigure committed
b713fba
G
[CI] Avoid full history fetch in skip-skill check (#79664)
Gu Shiwei committed
0dfb785
F
[Operator Mechanism] Make view dtype forward only (#79654)
feixi committed
5dd4894
B
align compat max grad with torch (#79657)
Bohao Fan committed
73245a1
B
align sum && mean on cpu with torch in single thread (#79568)
Bohao Fan committed
f984866
R
[Build] Fix missing dependencies for fused CUDA tests (#79585)
Runming Xie committed
df43822
August 14, 2026
F
[Operator Mechanism] Support baddbmm out_dtype for CUDA (#79638)
feixi committed
16037ff
August 13, 2026
W
[API Compatibility] Align paddle.Tensor.is_sparse, paddle.Tensor.type, paddle.Tensor.size api, paddle.nn.PReLU and paddle.distributions.categorical.Categorical -part (#79550)
Wenfei (Charles) Qi committed
4ab707b
F
[Operator Mechanism] Add CUDA stride-1 index gradient kernel (#79648)
feixi committed
4f5da6b
G
[HIP] hip support c++ 20 (#79646)
gouzil committed
0684083
D
[VMM] Add GPU memory history recorder with Python stack capture (#79609)
Difer committed
fe03d43
August 12, 2026
L
[Flex_CheckPoint]: Support high-dim-reshard for flex_ckpt (#79612)
Liumengyuan committed
fbb2493
F
[Operator Mechanism] Support addmm out_dtype for CUDA (#79613)
feixi committed
76d6cef
G
[CI] Update DCU image toolchain to GCC 12.2 (#79600)
gouzil committed
4dd5d82
August 11, 2026
R
[Environment Adaptation] Upgrade NCCL to 2.30.7 (#79643)
risemeup1 committed
9401c5b
G
[CI] Fix Doc-Preview GCC 11 runtime compatibility (#79617)
gouzil committed
f8ae875