issues
search
f-dangel
/
backpack
BackPACK - a backpropagation package built on top of PyTorch which efficiently computes quantities other than the gradient.
https://backpack.pt/
MIT License
548
stars
55
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Request: Extension for GroupNorm
#328
ParthS007
opened
1 month ago
1
Feature request: `BatchL2Grad` for `LayerNorm`
#327
f-dangel
opened
1 month ago
0
BackPACK with simple attention and additional layers
#326
nhianK
opened
2 months ago
1
Second Order Extensions for Custom Loss Modules
#325
JRopes
closed
3 months ago
8
[DEL] Remove warnings about a fixed bug in adaptive average 3d pooling
#324
f-dangel
opened
4 months ago
0
[FIX | FMT] RTD build, apply latest `black` and `isort`
#323
f-dangel
closed
4 months ago
0
Missing Support for BatchNorm and AdaptiveAvgPool in HBP methods (KFAC, KFRA, KFLR) and GGNMP
#322
satnavpt
opened
4 months ago
8
Extend backpack to deal with weighted sums
#321
rm-wu
opened
4 months ago
14
[Docs] Second-order Extension Example
#320
RobertDeibel
closed
4 months ago
0
Feature for backpack on VAEs
#319
shubham0704
opened
8 months ago
2
Facing error while Using DiagHessian for torchvision.models.resnet18
#318
AliGhadirii
closed
9 months ago
1
fix: ScaleModule and SumModule for DiagHessian.
#317
hlzl
opened
10 months ago
3
Missing implementation of supported layers for DiagHessian and BatchDiagHessian
#316
hlzl
opened
10 months ago
0
[CI] Drop `system_packages` in RTD config
#315
f-dangel
closed
10 months ago
0
Adaptive pooling for diagh
#314
fKunstner
closed
10 months ago
0
AdaptiveAvgPool not supported for 2nd order derivatives?
#313
hlzl
closed
10 months ago
4
Merge `1.6.0` back into `development`
#312
f-dangel
closed
12 months ago
0
[DOC] More changelog updates for 1.6
#311
f-dangel
closed
12 months ago
0
torch version < 2.x in `setup.cfg`
#310
elcorto
closed
12 months ago
2
Update master with code for `1.6.0` release
#309
f-dangel
closed
12 months ago
0
[DOC] Update changelog, prepare `1.6.0` release
#308
f-dangel
closed
1 year ago
0
[REQ] Support PyTorch 2.x
#307
f-dangel
closed
1 year ago
0
Container modules with advanced control flow & modules with multiple inputs
#306
m-lyon
opened
1 year ago
23
Second order extension
#305
yuanjiechen
closed
1 year ago
2
Encountered node that may break second-order extensions
#304
NizarIslah
closed
1 year ago
2
add support for torch 2.0?
#303
yahui624
closed
1 year ago
7
[DOC] Use case for `retain_graph`
#302
f-dangel
closed
1 year ago
0
[CI] Update code to latest `black` and `flake8`
#301
f-dangel
closed
1 year ago
0
Optimizing the locations of the Jacobians
#300
Ludvins
opened
1 year ago
5
Are customized loss functions supported?
#299
zzpustc
closed
3 months ago
10
Fix examples website
#298
fKunstner
closed
1 year ago
1
[DOC] Update changelog (prepare release of `1.5.2`)
#297
f-dangel
closed
1 year ago
0
[FIX] Support `torch>=1.13`
#296
f-dangel
closed
1 year ago
0
[FIX] Support for `torch>=1.13`
#295
f-dangel
closed
1 year ago
0
cannot import backpack nor extend
#294
fredguth
closed
1 year ago
9
[Feature Request] Levenberg Marquardt
#293
mizu-bai
opened
1 year ago
1
[ADD] Kronecker Hessian approximations for `ConvTranspose{1,2,3}d`
#292
f-dangel
closed
1 year ago
0
[ADD] Support Kronecker Hessian approximations for `Conv{1,2,3}d`
#291
f-dangel
closed
1 year ago
0
[DOC] Fully document `HBPConv2d`
#290
f-dangel
closed
1 year ago
0
[REF] Make output shape of `unfold_by_conv_transpose` consistent with convolution case
#289
f-dangel
closed
1 year ago
0
[DOC] Fully document transpose convolution utilities
#288
f-dangel
closed
1 year ago
0
[REF] Use `unfoldNd` package for input unfolding of transpose convolutions
#287
f-dangel
closed
1 year ago
0
[REF] Fully document convolution utilities
#286
f-dangel
closed
1 year ago
0
[REF] Use `unfoldNd` package for input unfolding of convolutions
#285
f-dangel
closed
1 year ago
0
[ADD] Test support for `BCEWithLogitsLoss` in `KFLR`
#284
f-dangel
closed
1 year ago
0
[ADD] Support `BCEWithLogitsLoss` in `KFAC`
#283
f-dangel
closed
1 year ago
0
[ADD] Support `BCEWithLogitsLoss` in `SqrtGGN{Exact, MC}`
#282
f-dangel
closed
1 year ago
0
Extending `BCEWithLogitsLoss` to non-binary labels
#281
f-dangel
opened
1 year ago
0
[ADD] Support `BCEWithLogitsLoss` in `(Batch)DiagGGN{Exact, MC}`
#280
f-dangel
closed
1 year ago
0
[ADD] Support `BCEWithLogitsLoss` in `(Batch)DiagHessian`
#279
f-dangel
closed
1 year ago
0
Next