Fix: Correct type hint for _compute_concentration_and_inv_freq #223
Conversation
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
| import math | ||
| import os | ||
| from dataclasses import dataclass | ||
| from typing import tuple |
There was a problem hiding this comment.
from typing import tuple will raise an ImportError when this module is imported because typing only exposes Tuple (capitalized) in Python 3.12+. The return annotation already uses the built‑in tuple[...] and does not require any import, so this new line makes the package fail to import entirely.
Useful? React with 👍 / 👎.
1 participant
Hello! This PR fixes an incorrect type hint in RotaryEmbedding._compute_concentration_and_inv_freq.
The function returns a tuple[float, torch.Tensor], but the type hint was incorrectly specified as -> torch.Tensor:.
This change ensures the signature is correct, which will prevent errors with static type checkers like mypy and improves overall code clarity.