aleximmer Laplace issues

aleximmer / Laplace

Laplace approximations for Deep Learning.

https://aleximmer.github.io/Laplace

MIT License

436 stars 63 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Help with Kronecker-factored Laplace approximation for ResNet

#210 joemrt opened 20 hours ago
2
Update installation instruction

#209 wiseodd closed 3 days ago
0
v0.2

#208 wiseodd closed 3 days ago
0
Add backend 'flowchart'

#207 wiseodd closed 3 days ago
0
Generic wrapper for Huggingface transformers

#206 wiseodd opened 3 days ago
0
Add some checks in `optimize_prior_precision`

#205 wiseodd closed 3 days ago
1
Update huggingface_example.md

#204 wiseodd closed 5 days ago
0
Efficient, universal, standalone Jacobian backend

#203 wiseodd opened 6 days ago
2
Caveats

#202 wiseodd closed 3 days ago
0
Implement KronLLLaplace fast functional variance

#201 wiseodd opened 1 week ago
0
Refactor reward-modeling likelihood

#200 runame closed 2 weeks ago
0
Remove `try-except` from gridsearch

#199 wiseodd closed 2 weeks ago
0
Add "Contributing" to the README

#198 wiseodd closed 3 weeks ago
0
Remove depreciated `loss_average` argument of `KFACLinearOperator` and add makefile for ruff

#197 runame closed 4 weeks ago
1
Add "Contributing" section to README

#196 wiseodd closed 3 weeks ago
3
FeatureExtractor and Python <3.11

#195 Ludvins closed 4 weeks ago
1
RunningNLLMetric returns infinite when running optimize_prior_precision

#194 MJordahn closed 2 weeks ago
3
Fit function of FullLLLaplace runs slower in Github version in comparison to pip version

#193 MJordahn closed 3 days ago
8
Functional Laplace Updated

#192 Ludvins opened 1 month ago
6
Mismatching pip installation and website documentation?

#191 MJordahn closed 1 month ago
5
fix typo in Jacobian dimension

#190 ruili-pml closed 1 month ago
0
subnetwork with Kronecker covariance

#189 ruili-pml closed 1 month ago
6
Merge changes from Metodj/Functional-Laplace

#188 Ludvins closed 1 month ago
1
Use backend-native Jacobians if available

#187 wiseodd closed 3 weeks ago
1
`torch.func` jacobian OOM

#186 wiseodd closed 3 days ago
2
Add more test coverage to the subset of params functionality

#185 wiseodd closed 4 weeks ago
0
Update docs

#184 wiseodd closed 3 weeks ago
1
Move back to ASDL's main repo as dependency

#183 aleximmer closed 2 months ago
0
Make `CurvatureInterface.functorch_jacobians` simply an alias to `CurvatureInterface.jacobians`

#182 wiseodd closed 3 weeks ago
1
Prior precision and prior mean of `LLLaplace` are hardcoded

#181 wiseodd closed 2 months ago
2
Typehinting

#180 wiseodd closed 3 weeks ago
2
Fix all linting issues and format all files with `ruff`

#179 wiseodd closed 4 weeks ago
0
Handle logits with more than 2 axes in `torch.func` diag and full backends

#178 wiseodd opened 2 months ago
0
Replace hardcoded class index with `logit_class_dim` argument

#177 wiseodd opened 2 months ago
7
Add `ruff check`, `ruff format --check`, and `pytest` Github actions

#176 wiseodd closed 4 weeks ago
1
Fast `KronLLLaplace` joint predictive

#175 wiseodd opened 2 months ago
0
Support for structured Kronecker factors

#174 wiseodd opened 2 months ago
1
Prevent computing posterior precision in KronLaplace when it's not fitted

#173 wiseodd closed 4 weeks ago
0
Add an option to reduce LLM features in `LLLaplace`

#172 wiseodd closed 4 weeks ago
0
Adding prior precision in `KronLaplace` is broken

#171 wiseodd closed 4 weeks ago
0
Feature caching mechanism in LLLA

#170 wiseodd closed 4 weeks ago
0
Handle feature reduction properly in LLLA

#169 wiseodd closed 4 weeks ago
0
Make the dict keys for models with dict-like inputs general

#168 wiseodd closed 1 month ago
1
Doc & example for reward modeling

#167 wiseodd closed 1 month ago
1
Add example & doc for reward modeling functionality

#166 wiseodd closed 1 month ago
0
Generalized dict-input mechanism

#165 wiseodd closed 1 month ago
0
Cache features in LLLA with MC predictive

#164 wiseodd closed 2 months ago
1
Conform to PyTorch convention in the loss

#163 wiseodd opened 2 months ago
1
Support for `torchtune`'s models and datasets

#162 wiseodd opened 2 months ago
0
Efficient LLLA predictive samples

#161 wiseodd closed 4 weeks ago
0