ApolloResearch rib issues

ApolloResearch / rib

Library for methods related to the Local Interaction Basis (LIB)

MIT License

2 stars 0 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Implement paper plots feedback

#367 stefan-apollo closed 6 months ago
0
Paper plotting scripts

#366 stefan-apollo closed 6 months ago
0
Error in distributed test

#365 danbraunai-apollo opened 7 months ago
0
Preparation for publication

#364 danbraunai-apollo closed 7 months ago
2
Added new hook & function for feature viz tool

#363 stefan-apollo closed 8 months ago
0
[Not a PR] Play/feature viz dashboard

#362 stefan-apollo closed 8 months ago
0
Refactor of plotting code to merge modularity and labels, plus a bunch of small improvements

#361 stefan-apollo closed 8 months ago
0
Update the modular addition config file

#360 stefan-apollo closed 8 months ago
0
Improved docstring / comments

#359 stefan-apollo closed 8 months ago
0
WIP Updates to modularity PR

#358 stefan-apollo closed 8 months ago
0
Improve ablations

#357 nix-apollo closed 8 months ago
1
Set huggingface cache at the right point

#356 stefan-apollo closed 8 months ago
0
Take cache mode activations off the GPU

#355 stefan-apollo closed 8 months ago
0
`create_hf_dataset` optional argument `model_n_ctx` is not optional

#354 stefan-apollo opened 8 months ago
0
WIP Store commit ID in RIB builds

#353 stefan-apollo closed 6 months ago
2
naive_gradient_flow and basis_formula=svd should confict

#352 stefan-apollo opened 8 months ago
0
Fix for plotting bug introduced in #348

#351 stefan-apollo closed 8 months ago
1
Use a huggingface cache

#350 stefan-apollo closed 8 months ago
1
Modularity Analysis

#349 nix-apollo closed 6 months ago
2
Feature/plotting improvements

#348 nix-apollo closed 8 months ago
0
Ablations with reference to baseline

#347 stefan-apollo closed 8 months ago
2
Fix/negative variance edge ablation [Replacement for #323]

#346 stefan-apollo closed 8 months ago
1
Isolate the variance in layernorm into a separate RIB dir

#345 nix-apollo closed 8 months ago
1
Use separate (larger) dataset for gram (and mean) matrices

#344 stefan-apollo closed 8 months ago
4
[Not a PR] play/ngf tinystories

#343 stefan-apollo closed 7 months ago
0
Switch roneneldan/TinyStories -> skeskinen/TinyStories-hf

#342 danbraunai-apollo closed 8 months ago
1
Implementation of PCA weighted by the size of the gradient on each datapoint

#341 jakeapollo closed 8 months ago
1
Bugfix for distributed seeds

#340 nix-apollo closed 8 months ago
2
Some tests fail on A100 (that do not fail on A6000)

#339 stefan-apollo opened 8 months ago
0
Nix' magic memory saving changes

#338 stefan-apollo closed 8 months ago
3
WIP Feature/modularity plot

#337 stefan-apollo closed 7 months ago
1
Existing outfiles (check_outfile_overwrite = False) is handled inconsistently

#336 stefan-apollo opened 9 months ago
0
load_interaction_rotations should return calc_Cs_time

#335 stefan-apollo opened 9 months ago
0
Ignore 0th position

#334 nix-apollo closed 6 months ago
2
Naive implementation of gradient flow (in refactored code)

#333 stefan-apollo closed 8 months ago
8
Distributed calculation for basis

#332 nix-apollo closed 9 months ago
0
WIP: Adds ability to set off-diagonal weights in modular mlp

#331 LuciusApollo closed 6 months ago
1
Build test fixes

#330 nix-apollo closed 9 months ago
0
Fix flaky test `test_pythia_14m_build_graph_jacobian_stochastic`

#329 nix-apollo closed 9 months ago
0
testing

#328 danbraunai-apollo closed 9 months ago
0
Updates for python 3.12

#327 nix-apollo closed 9 months ago
2
Make workflow independent of host

#326 danbraunai-apollo closed 9 months ago
0
Specify integral method by layer

#325 nix-apollo closed 9 months ago
0
Make RIB more memory efficient

#324 nix-apollo opened 9 months ago
0
Fix/negative variance edge ablation

#323 stefan-apollo closed 8 months ago
5
[WIP] Feature/improve defaults and yamls

#322 stefan-apollo closed 6 months ago
2
WIP Feature/integral depending on module

#321 stefan-apollo closed 9 months ago
1
Cache transformers datasets

#320 danbraunai-apollo opened 9 months ago
1
Allow distributed edge splitting over out_dim

#319 danbraunai-apollo closed 9 months ago
0
Negative variance iff edge ablations in Split LN

#318 stefan-apollo opened 9 months ago
0