-
Notifications
You must be signed in to change notification settings - Fork 4.5k
Pull requests: Tencent/ncnn
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
pnnx: support PyTorch ExportedProgram .pt2 models
pnnx
tool
#6959
opened Sep 4, 2026 by
geekgoer
Loading…
fix(pnnx): widen bool MemoryData attributes to fp32 in .bin (#6855)
pnnx
tool
#6956
opened Sep 3, 2026 by
ktz03
Contributor
Loading…
3 tasks
fix(mat): zero-init freshly allocated Mat buffers (#6765)
core
#6954
opened Sep 3, 2026 by
ktz03
Contributor
Loading…
2 tasks
fix(vulkan): omit zero-count descriptor pool sizes
core
#6952
opened Sep 3, 2026 by
ktz03
Contributor
Loading…
2 tasks
fix(inversespectrogram): validate n_fft/winlen in load_param
layer
#6950
opened Sep 2, 2026 by
ktz03
Contributor
Loading…
2 tasks
fix(net): bounds-check blob indexes in load_param_bin
core
#6949
opened Sep 2, 2026 by
ktz03
Contributor
Loading…
2 tasks
fix(spectrogram): validate n_fft/winlen in load_param
layer
#6948
opened Sep 2, 2026 by
ktz03
Contributor
Loading…
2 tasks
x86 avx-ne-convert optimization for bf16 gemm and conversion
x86
#6947
opened Sep 2, 2026 by
nihui
Member
Loading…
pnnx for torch exported program
pnnx
tool
#6946
opened Sep 2, 2026 by
AtomAlpaca
Contributor
Loading…
pnnx for torch exported program
layer
pnnx
tool
x86
#6941
opened Sep 1, 2026 by
Edwardssss
Contributor
Loading…
pnnx: preserve batch semantics across reshape and flatten
pnnx
tool
#6934
opened Aug 29, 2026 by
futz12
Contributor
Loading…
vulkan: apply atan2 zero workaround to packed shaders
vulkan
#6930
opened Aug 25, 2026 by
LEKKO2310
Loading…
fix(multiheadattention): add num_heads and embed_dim validation to prevent FPE
layer
#6928
opened Aug 23, 2026 by
lxcxjxhx
Loading…
fix: stabilize logaddexp expression evaluation
core
test
#6924
opened Aug 20, 2026 by
primorLee
Loading…
fix unvalidated layer counts, out-of-bounds blob index and division by zero in load_param and load_param_bin
core
#6922
opened Aug 18, 2026 by
kaltsit-10
Contributor
Loading…
Fix empty string bounds check in ParamDict::load_param
core
#6918
opened Aug 17, 2026 by
beilzx
Loading…
fix(paramdict): add non-negative check for array length in load_param
core
#6917
opened Aug 17, 2026 by
beilzx
Loading…
fix(net): add blob_index bounds check in load_param
core
#6916
opened Aug 17, 2026 by
beilzx
Loading…
fix(convolutiondepthwise): add group == 0 guard to prevent FPE
layer
#6912
opened Aug 17, 2026 by
beilzx
Loading…
Previous Next
ProTip!
no:milestone will show everything without a milestone.