Skip to content

minor modify #2386

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

Merged
merged 2 commits into from
Sep 10, 2025
+5 −2
Merged

minor modify #2386

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
374617e
minor modify
pass-lin Sep 6, 2025
a5bdcc3
fix
pass-lin Sep 6, 2025
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
3 changes: 2 additions & 1 deletion keras_hub/src/models/backbone.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -130,7 +130,8 @@ def from_preset(
1. a built-in preset identifier like `'bert_base_en'`
2. a Kaggle Models handle like `'kaggle://user/bert/keras/bert_base_en'`
3. a Hugging Face handle like `'hf://user/bert_base_en'`
4. a path to a local preset directory like `'./bert_base_en'`
4. a ModelScope Face handle like `'modelscope://user/bert_base_en'`
5. a path to a local preset directory like `'./bert_base_en'`
Copy link
Contributor

@gemini-code-assist gemini-code-assist bot Sep 6, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

There's a small typo in the description for the ModelScope handle. It should be ModelScope handle instead of ModelScope Face handle for consistency with other parts of the codebase (e.g., keras_hub/src/utils/preset_utils.py) and to avoid confusion with Hugging Face.

Suggested change
4. a ModelScope Face handle like `'modelscope://user/bert_base_en'`
5. a path to a local preset directory like `'./bert_base_en'`
4. a ModelScope handle like `'modelscope://user/bert_base_en'`
5. a path to a local preset directory like `'./bert_base_en'`


This constructor can be called in one of two ways. Either from the base
class like `keras_hub.models.Backbone.from_preset()`, or from
Expand Down
4 changes: 3 additions & 1 deletion keras_hub/src/models/sam/sam_prompt_encoder.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -67,7 +67,9 @@ def __init__(
self.activation = activation

self.positional_embedding_layer = RandomFrequencyPositionalEmbeddings(
num_positional_features=self.hidden_size // 2, scale=1
num_positional_features=self.hidden_size // 2,
scale=1,
dtype=self.dtype,
)
Comment on lines 69 to 73
Copy link
Contributor

@gemini-code-assist gemini-code-assist bot Sep 6, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

high

Passing dtype=self.dtype here is a good step towards supporting bfloat16 inference. However, to ensure full compatibility and prevent potential dtype mismatches, this change should be applied consistently to all other layers instantiated within the __init__ method.

For example, the keras.layers.Embedding layers and the layers inside mask_downscaler are also missing the dtype argument. Without this, the model might not function correctly when using bfloat16 precision.

Copy link
Contributor Author

@pass-lin pass-lin Sep 6, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I have tested it, just need to modify it here.


self.foreground_point_embed = keras.layers.Embedding(
Expand Down
Loading