Trainer error handling

[Mandolaro]

📝 Summary

• 1. Improve Error handling

✅ Checklist

• Docs updated (if necessary)
• Changelogs are updated (if necessary)

[gemini-code-assist]

Summary of Changes

Hello @Mandolaro, 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 a significant new capability for model training by integrating Reinforcement Learning Fine-Tuning (RFT) with Fireworks AI. It adds a dedicated trainer module to manage the iterative training process, including model deployment, snapshot management, and reward-based learning. Concurrently, the tracing system has been upgraded to seamlessly monitor these new trainable models and extract conversational history for training. The changes also involve a cleanup and simplification of existing evaluation APIs, removing older trace-based evaluation methods, and minor refactorings to the custom scorer upload functionality, improving overall system robustness and clarity.

Highlights

• Reinforcement Learning Fine-Tuning (RFT) Integration: A comprehensive Reinforcement Learning Fine-Tuning (RFT) framework has been introduced, leveraging Fireworks AI, to enable iterative model training and improvement within the platform.
• New trainer Module: A new trainer module has been added, comprising JudgmentTrainer for orchestrating the RFT process, TrainableModel for managing model deployments and snapshots, and TrainerConfig/ModelConfig for configuration and state persistence.
• Enhanced Tracing Capabilities: The Tracer component has been enhanced to support wrapping TrainableModel instances, extract message history from traces for training data, and offers a new show_trace_urls option for controlling the visibility of UI links.
• Streamlined Evaluation APIs: The older run_trace_evaluation methods have been deprecated and removed from the JudgmentClient and run_evaluation.py, simplifying and streamlining the overall evaluation interface.
• Custom Scorer Upload Refinements: The CLI command and internal API endpoint for uploading custom scorers have been renamed from 'save' to 'upload', along with updates to the corresponding API path, for clearer terminology.

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 in your pull request via creating an issue comment (i.e. comment on the pull request page) using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands.

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 issue 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 introduces comprehensive error handling throughout the trainer logic by wrapping method calls in try...except blocks and raising a custom JudgmentAPIError. This is a great improvement for robustness and provides more informative errors to the user. The changes are generally well-implemented. I have one suggestion to refactor an exception handling block to be more robust by relying on exception types instead of string matching on the error message.

[JCamyre]

LGTM! Sick.