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

Jump to bottom

Add documentation and error messages for w_init and w_init_scale to avoid confusion #541

Merged
merged 1 commit into from
Oct 18, 2022
Merged

Add documentation and error messages for w_init and w_init_scale to avoid confusion #541

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
8 changes: 6 additions & 2 deletions haiku/_src/attention.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -71,7 +71,9 @@ def __init__(
num_heads: Number of independent attention heads (H).
key_size: The size of keys (K) and queries used for attention.
w_init_scale: DEPRECATED. Please use w_init instead.
w_init: Initialiser for weights in the linear map.
w_init: Initialiser for weights in the linear map. Once `w_init_scale` is
fully deprecated `w_init` will become mandatory. Until then it has a
default value of `None` for backwards compatability.
value_size: Optional size of the value projection (V). If None, defaults
to the key size (K).
model_size: Optional size of the output embedding (D'). If None, defaults
Expand All @@ -92,7 +94,9 @@ def __init__(
if w_init and w_init_scale:
raise ValueError("Please provide only `w_init`, not `w_init_scale`.")
if w_init is None and w_init_scale is None:
raise ValueError("Please provide a weight initializer: `w_init`.")
raise ValueError("Please provide a weight initializer: `w_init`. "
"`w_init` will become mandatory once `w_init_scale` is "
"fully deprecated.")
if w_init is None:
w_init = hk.initializers.VarianceScaling(w_init_scale)
self.w_init = w_init
Expand Down