fix: update rag example

[e06084]

No description provided.

[gemini-code-assist]

Summary of Changes

Hello @e06084, 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 enhances the robustness of RAG evaluation metrics by preventing crashes when essential fields like 'answer' or 'expected_output' are missing. Instead, it assigns a zero score with specific diagnostic labels, making the evaluation process more resilient. Additionally, the RAG example configuration has been updated to utilize a new embedding model and introduces a more organized structure for embedding service parameters.

Highlights

• Robust RAG Evaluation: Modified RAG evaluation metrics (Answer Relevancy, Context Precision, Context Recall) to gracefully handle missing 'answer' or 'expected_output' fields by returning a score of 0 instead of raising an error, improving the stability of evaluations.
• Updated Embedding Model: The default embedding model in the RAG example has been updated from 'text-embedding-3-large' to 'doubao-embedding-large-text-250515'.
• Refactored Embedding Configuration: The RAG example now uses a dedicated 'embedding_config' dictionary within 'llm_config_embedding' for better structuring of embedding service parameters, enhancing clarity and flexibility.

Using Gemini Code Assist

The 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 /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

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 .gemini/ folder in the base of the repository. Detailed instructions can be found here.

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

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request refactors the RAG evaluators to gracefully handle cases where the answer or expected_output is missing, returning a score of 0 instead of raising an error. This improves the robustness of the evaluation pipeline. The configuration for embedding models has also been updated for better clarity, and the corresponding example file has been adjusted. My review focuses on improving code maintainability by addressing code duplication. The logic for handling empty inputs is repeated across three different evaluator classes. I've suggested extracting this into a shared helper method. I also pointed out some minor data fetching redundancy.

[gemini-code-assist]

There's significant code duplication in handling empty answer fields across multiple RAG evaluator classes. This same logic block is present in llm_rag_answer_relevancy.py (lines 224-234) and a similar one for expected_output in llm_rag_context_recall.py (lines 237-252).

To improve maintainability and adhere to the DRY principle, consider extracting this logic into a protected helper method in the BaseOpenAI class. This method could handle checking for empty inputs and generating the standard 0-score EvalDetail response.

For example, a helper method could be added to BaseOpenAI:

from typing import Optional
...
@classmethod
def _handle_empty_input(cls, value: str, field_name: str, metric_label: str, reason_msg: str) -> Optional[EvalDetail]:
    if not value:
        log.warning(f"{cls.__name__} 评估: {field_name} 字段为空，直接返回 0 分")
        result = EvalDetail(metric=cls.__name__)
        result.score = 0.0
        result.status = True
        result.label = [metric_label]
        result.reason = [reason_msg]
        return result
    return None

You could then use it in the eval methods like this:

# In LLMRAGContextPrecision.eval
raw_data = getattr(input_data, 'raw_data', {})
answer = input_data.content or raw_data.get("answer", "")
if empty_result := cls._handle_empty_input(
    answer,
    "answer",
    "QUALITY_BAD.CONTEXT_PRECISION_NO_ANSWER",
    "answer 字段为空，无法评估上下文精度，分数设为 0"
):
    return empty_result

This would make the code cleaner and easier to manage.

[gemini-code-assist]

The raw_data and answer variables are now being fetched both here in the eval method and again in the build_messages method (lines 163-165), which is called later in this method. This introduces some redundancy.

While the performance impact is likely negligible, to improve code clarity and reduce redundancy, you could consider fetching these values only once in eval and then passing them to build_messages. This would require changing the signature of build_messages to accept answer and question as arguments. This pattern of redundant data fetching is also present in llm_rag_answer_relevancy.py and llm_rag_context_recall.py.