Single SAEs trained on the residual stream activation vectors from every transformer layer simultaneously: https://arxiv.org/abs/2409.04185
Tim Lawson
tim-lawson
AI & ML interests
Mechanistic interpretability, language modelling, semantics
Recent Activity
updated
a dataset
2 days ago
tim-lawson/mlsae-gpt2-x64-k32-dists
updated
a model
2 days ago
tim-lawson/mlsae-gpt2-x64-k32
updated
a model
2 days ago
tim-lawson/mlsae-gpt2-x64-k32-tfm
Organizations
None yet
models
165
tim-lawson/mlsae-gpt2-x64-k32
Updated
•
10
tim-lawson/mlsae-gpt2-x64-k32-tfm
Updated
•
44
tim-lawson/sae-gpt2-x64-k32-layers-01234567891011
Updated
•
4
tim-lawson/sae-gpt2-x64-k32-tfm-layers-01234567891011
Updated
•
5
tim-lawson/mlsae-pythia-160m-deduped-x64-k32-tfm
Updated
•
184
tim-lawson/mlsae-pythia-70m-deduped-x64-k32-tfm
Updated
•
229
tim-lawson/sae-pythia-1.4b-deduped-x64-k32-layers-01234567891011121314151617181920212223
Updated
•
2
tim-lawson/sae-pythia-1.4b-deduped-x64-k32-tfm-layers-01234567891011121314151617181920212223
Updated
•
2
tim-lawson/sae-pythia-410m-deduped-x64-k32-layers-0
Updated
•
2
tim-lawson/sae-pythia-410m-deduped-x64-k32-tfm-layers-0
Updated
•
2
datasets
58
tim-lawson/mlsae-gpt2-x64-k32-dists
Preview
•
Updated
•
7
tim-lawson/sae-pythia-160m-deduped-x64-k32-tfm-layers-11-dists
Viewer
•
Updated
•
49.2k
•
8
tim-lawson/sae-pythia-160m-deduped-x64-k32-tfm-layers-10-dists
Viewer
•
Updated
•
49.2k
•
8
tim-lawson/sae-pythia-160m-deduped-x64-k32-tfm-layers-8-dists
Viewer
•
Updated
•
49.2k
•
7
tim-lawson/sae-pythia-160m-deduped-x64-k32-tfm-layers-9-dists
Viewer
•
Updated
•
49.2k
•
7
tim-lawson/sae-pythia-160m-deduped-x64-k32-tfm-layers-7-dists
Viewer
•
Updated
•
49.2k
•
7
tim-lawson/sae-pythia-160m-deduped-x64-k32-tfm-layers-5-dists
Viewer
•
Updated
•
49.2k
•
7
tim-lawson/sae-pythia-160m-deduped-x64-k32-tfm-layers-6-dists
Viewer
•
Updated
•
49.2k
•
7
tim-lawson/sae-pythia-160m-deduped-x64-k32-tfm-layers-2-dists
Viewer
•
Updated
•
49.2k
•
7
tim-lawson/sae-pythia-160m-deduped-x64-k32-tfm-layers-0-dists
Viewer
•
Updated
•
49.2k
•
7