Revision of prompts for smaller models

[jpodivin]

Our new prompts have exhibited irregular behavior when used with smaller models, especially with limited context sizes.

By making small modifications to them, we can ensure that the worst aspects, such as excessive anchoring in examples and verbosity, is avoided.

[gemini-code-assist]

Summary of Changes

Hello @jpodivin, 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 introduces targeted revisions to two core system prompts, snippet_system_prompt.j2 and staged_system_prompt.j2, to enhance their compatibility and effectiveness with smaller language models. The primary goal is to streamline the prompts by removing unnecessary instructions and verbose phrasing, thereby enabling models to deliver more focused, accurate, and concise analyses of RPM package build failures, particularly in resource-constrained environments.

Highlights

• Prompt Optimization for Smaller Models: Revised system prompts to improve performance and reduce irregular behavior when used with smaller language models, especially those with limited context sizes.
• Reduced Verbosity and Anchoring: Eliminated excessive verbosity and anchoring in examples within the prompts to make them more efficient and focused.
• Refined snippet_system_prompt.j2: Updated the prompt to be more concise, focusing on providing short explanations of diagnostic information and explicitly instructing the model to avoid quoting raw snippets or fabricating information.
• Enhanced staged_system_prompt.j2: Modified this prompt to guide the model in identifying primary errors, ignoring generic messages, and providing specific, concise root cause explanations and solutions without boilerplate.

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 system prompts to be more concise and direct, which is a good step towards improving their performance on smaller language models. The updated instructions are clearer and more specific, which should lead to more consistent and useful outputs. I've included one suggestion to address a minor inconsistency in the persona description across the different prompt files.

[gemini-code-assist]

This prompt now specifies the "Fedora ecosystem", while the unchanged system_prompt.j2 refers to the "RHEL ecosystem", and the modified snippet_system_prompt.j2 removes the ecosystem mention altogether. This inconsistency could lead to confusion or incorrect context for the model. For better consistency and generality, consider making this persona description similar to the one in snippet_system_prompt.j2, which removes the specific ecosystem and combines the role and purpose into a single line.

You are an expert system specialized in RPM package build failures. Your purpose is to diagnose the root cause of failures and propose solutions.