issues
search
adamkarvonen
/
SAEBench
7
stars
10
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
don't fail entire job if one SAE fails
#41
hijohnnylin
opened
3 days ago
0
dpm
#40
hijohnnylin
closed
3 days ago
0
add gemma-2-9b default DTYPE and BATCH_SIZE
#39
hijohnnylin
closed
3 days ago
0
model_name arg in core seems to do nothing
#38
hijohnnylin
opened
4 days ago
1
Core Eval (minor?) Issues
#37
hijohnnylin
closed
4 days ago
3
Ravel Uniprobe
#36
canrager
opened
5 days ago
0
Add pca
#35
adamkarvonen
closed
5 days ago
0
fixing excessively low precision
#34
curt-tigges
closed
1 week ago
0
Add baselines
#33
adamkarvonen
closed
1 week ago
0
Added option to exclude special tokens from SAE reconstruction
#32
curt-tigges
closed
1 week ago
0
Activation consolidation
#31
adamkarvonen
closed
1 week ago
0
Mdl fixes
#30
adamkarvonen
closed
2 weeks ago
0
steps towards ravel in tlens
#29
amakelov
opened
2 weeks ago
0
Update unlearning output format
#28
hijohnnylin
closed
2 weeks ago
0
Update JSON schema filenames
#27
hijohnnylin
closed
2 weeks ago
0
Update ui_default_display, titles for display
#26
hijohnnylin
closed
2 weeks ago
0
added tests for core eval output
#25
curt-tigges
closed
2 weeks ago
0
New Core output format, plus converter
#24
hijohnnylin
closed
2 weeks ago
0
Unlearning adapt
#23
adamkarvonen
closed
3 weeks ago
0
handle case where gated SAEs don't have b_enc
#22
curt-tigges
closed
3 weeks ago
0
Shift sparse probing descriptions
#21
adamkarvonen
closed
3 weeks ago
0
fix: eval_result_unstructured should be optional
#20
hijohnnylin
closed
3 weeks ago
0
Core eval incremental saving
#19
curt-tigges
closed
3 weeks ago
0
set k = 1, 2, 5 default display = true for sparse probing
#18
hijohnnylin
closed
3 weeks ago
0
Feature: Support unstructured eval output
#17
hijohnnylin
closed
4 weeks ago
0
Added core evals to repo
#16
curt-tigges
closed
4 weeks ago
0
Use Pydantic for eval configs and outputs for annotations and portability
#15
hijohnnylin
closed
4 weeks ago
0
Shift sparse probing updates
#14
adamkarvonen
closed
4 weeks ago
0
Minor shift improvements
#13
adamkarvonen
closed
1 month ago
0
Demo of Changes to enable easy running of evals at scale (using absorption)
#12
jbloomAus
closed
1 month ago
11
Add additional sparse probing datasets
#11
adamkarvonen
closed
1 month ago
0
Initial RAVEL code
#10
curt-tigges
closed
1 month ago
0
Unlearning cleanup
#9
adamkarvonen
closed
1 month ago
0
Autointerp eval
#8
callummcdougall
closed
1 week ago
6
implement unlearning eval
#7
yeutong
closed
1 month ago
1
Implement MDL eval
#6
koayon
closed
2 weeks ago
1
Rename utils to avoid name conflict
#5
koayon
closed
1 month ago
4
Shift eval
#4
adamkarvonen
closed
1 month ago
0
Feature Absorption Eval
#3
chanind
closed
1 month ago
5
Sparse probing add datasets
#2
adamkarvonen
closed
1 month ago
0
Restructure
#1
canrager
closed
1 month ago
0