Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

fix(sae): support grid searching for best init #50

Merged
merged 1 commit into from
Aug 19, 2024
Merged

fix(sae): support grid searching for best init #50

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 4 additions & 3 deletions src/lm_saes/sae.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -332,6 +332,7 @@ def compute_loss(
| None
) = None,
return_aux_data: bool = True,
during_init: bool = False,
) -> Union[
Float[torch.Tensor, "batch"],
tuple[
Expand Down Expand Up @@ -373,7 +374,7 @@ def compute_loss(
if self.cfg.sparsity_include_decoder_norm:

l_l1 = torch.norm(
feature_acts_normed * self.decoder_norm(),
feature_acts_normed * self.decoder_norm(during_init=during_init),
p=self.cfg.lp,
dim=-1,
)
Expand Down Expand Up @@ -669,7 +670,7 @@ def grid_search_best_init_norm(search_range: List[float]) -> float:
test_sae.encoder.weight.data = (
test_sae.decoder.weight.data.T.clone().contiguous()
)
mse = test_sae.compute_loss(x=activation_in, label=activation_out)[1][0]["l_rec"].mean().item() # type: ignore
mse = test_sae.compute_loss(x=activation_in, label=activation_out, during_init=True)[1][0]["l_rec"].mean().item() # type: ignore
losses[norm] = mse
best_norm = min(losses, key=losses.get) # type: ignore
return best_norm
Expand All @@ -687,7 +688,7 @@ def grid_search_best_init_norm(search_range: List[float]) -> float:
)

test_sae.set_decoder_norm_to_fixed_norm(
best_norm_fine_grained, force_exact=True
best_norm_fine_grained, force_exact=True, during_init=True
)
test_sae.encoder.weight.data = (
test_sae.decoder.weight.data.T.clone().contiguous()
Expand Down