fix:readme.md(#204) #288
fix:readme.md(#204) #288
Conversation
Signed-off-by: hydrogenair <3238172875@qq.com>
kubeedge-bot
added
the
kind/documentation
Summary of ChangesHello @hydrogenair, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request significantly enhances the documentation for the Privacy-Aware Cloud-Edge Collaborative LLM Inference Framework. It aims to make the README more comprehensive and actionable by improving its structure, clarifying technical details, and providing more explicit guidance for benchmarking, ultimately helping users better understand and apply the framework. Highlights
Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here. You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension. Footnotes
|
kubeedge-bot
added
the
size/M
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: The full list of commands accepted by this bot can be found here. DetailsNeeds approval from an approver in each of these files:Approvers can indicate their approval by writing |
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request significantly improves the README documentation by adding a table of contents, numbering sections for better structure, and providing much more detailed explanations for the evaluation metrics. The changes make the documentation more comprehensive and easier to follow. I've left a few comments regarding minor formatting issues and a potentially unintentional content removal in the 'Future Work' section. Overall, this is a great update to the documentation.
| - Inputs: Entity annotations from original text and sanitized text from cloud model output | ||
| - Calculation: Compares detected sensitive entities in sanitized text with TAB benchmark annotations, using TAB's marked entity positions and categories to measure detection accuracy. | ||
| - **Sensitive Entity Leakage Score (SELS)**: | ||
| - Inputs: Entity annotations from the original text and unsanitized entities in the cloud model output | ||
| - Calculation: Weighted score based on entity types that should be anonymized but remain exposed: | ||
| - `DIRECT` identifiers (require anonymization): weight = 5 | ||
| - `QUASI` identifiers (require anonymization): weight = 3 | ||
| - `NO_MASK` entities (no anonymization needed): weight = 0 | ||
| - Computed using TAB's predefined classification weights. | ||
| - **Inference Attack Resistance (IAR)**: | ||
| - Inputs: Sanitized text from cloud model output and auxiliary information from cloud model results | ||
| - Calculation: | ||
| 1. Convert both sanitized text and auxiliary information to semantic vectors using a sentence embedding model | ||
| 2. Compute cosine similarity between vectors (higher similarity indicates greater inference vulnerability) | ||
| 3. Final IAR = 1 - cosine similarity | ||
| - **Contextual Privacy Preservation (CPP)**: | ||
| - Inputs: Original text and sanitized text from cloud model output | ||
| - Calculation: | ||
| 1. Convert both texts to semantic vectors using `SentenceTransformer('all-MiniLM-L6-v2')` | ||
| 2. Cosine similarity between vectors is used directly as CPP score |
There was a problem hiding this comment.
There are several trailing or extra whitespaces in this section that should be removed for cleaner markdown source code:
- Line 76: trailing space at the end.
- Line 79: trailing space at the end.
- Line 86: two trailing spaces at the end.
- Line 92: trailing space at the end.
- Line 95: two spaces before 'is'.
| - Enhance cross-framework compatibility for seamless integration with diverse cloud and edge platforms | ||
|
|
||
| - Develop automated tuning mechanisms for optimal privacy-performance tradeoff configurations | ||
| - Extend support for multi-modal data (images, audio) in collaborative inference scenarios No newline at end of file |
There was a problem hiding this comment.
This list item about extending support for multi-modal data seems to have been removed. Given that it's a valid direction for future work, was this removal intentional? If not, it should be restored.
| - Develop automated tuning mechanisms for optimal privacy-performance tradeoff configurations | ||
| - Extend support for multi-modal data (images, audio) in collaborative inference scenarios No newline at end of file | ||
|
|
||
|
No newline at end of file |
There was a problem hiding this comment.
This file ends with a line containing only spaces and no final newline character. It's a common convention to end text files with a single newline character. Please consider adding one.
2 participants
What type of PR is this?
/kind documentation
What this PR does / why we need it:
This PR refines and supplements the README documentation for the
Privacy-Aware Cloud-Edge Collaborative LLM Inference Framework with Adaptive Privacy Protection(originally submitted in PR #273 for issue #204). Key improvements include:These enhancements make the documentation more comprehensive and actionable, helping users quickly understand and apply the framework while aligning with the original requirements of issue #204.
Which issue(s) this PR fixes:
Fixes #204