Release 2.1.10

Version 2.1.10 Update
Critical new capabilities: Implemented EnsembleClassifier in ensemble.py (ensemble of AutoClassifier models with varying hyperparameters and random seeds), updated Ensemble with AutoClassifier.ipynb, and enhanced reports.py's generate_html_pipeline for accurate pipeline reporting.
No stability fixes — focused on robust model ensembles and actionable reporting (builds on 2.1.9’s pipeline reliability).

Release 2.1.9

Version 2.1.9 Update
Critical fixes: Resolved inconsistent action tensor shapes in MultiBanditNet for multi-option scenarios and stabilized OptionCriticEnv option selection logic.
No new features — targeted pipeline reliability for complex option environments (builds on 2.1.8’s PPO stability).

Release 2.2.2rc0

Version 2.2.2rc0 Release Notes:
This Pre-release iteration integrates prior fixes while incorporating targeted optimizations such as Rust-based implementations for performance gains across critical modules.

Enhancements include:

• Expanded environment compatibility: Refined documentation clarity, and additional feature support. Some components leverage lightweight Rust dummy code to accelerate specific workflows without introducing stability risks. Testing confirms minor adjustments remain stable pending final validation.

Further refinements are underway to ensure readiness for deployment.

Release 2.1.8

Version 2.1.8 Update:
This iteration resolves prior GPU inefficiencies and stabilizes PPO implementations through refined logic corrections while enhancing documentation clarity.

Key Enhancements Include:

• Expanded Environment Support: Improved pipeline functionality for scalability, and updated resource management optimizations derived from prior fixes.
• Continued Focus On Stability and Usability: Ensures seamless adoption across diverse use cases.

Release 2.1.7

Version 2.1.7 Update:
We are pleased to announce version 2.1.7 of our likelihood package, incorporating critical fixes and enhancements to solidify its reliability and utility.

Key Updates Include:

• Fixed Critical GPU Usage Bug: Resolution of prior issues impacting resource efficiency in training pipelines.
• Enhanced OptionCriticEnv Stability: Improved performance consistency across diverse environments.
• Docs Enhancements: Updated documentation to clarify workflows and usage guidelines.

Additional improvements reflect refined code quality, expanded example coverage, and greater flexibility in deployment scenarios. This iteration further strengthens the package’s role as a versatile tool for reinforcement learning applications.

Note: Changes also include cleanup of redundant modules and refined pipeline integrations.

This version prioritizes stability, documentation accuracy, and adaptability while addressing prior concerns.

Release 2.2.1rc0

Pre-Release 2.2.1rc0 Update:
We are releasing Pre-Release 2.2.1rc0 as a developmental tool to refine testing capabilities while maintaining focus on stability and clarity for development teams. This iteration builds on prior enhancements but remains under active review, with features finalized only in follow-up stages.

Key Additions:

• Rust Testing Integration: Early-stage integration of Rust-based test frameworks to accelerate validation without disrupting existing workflows.
• Development Focus: Optimizations tailored for iterative prototyping and rapid feedback loops.
• Clarified Scope: Changes remain experimental, with full implementation pending finalization in 2.2.1.

Note: This version is intended solely for development teams to assess enhancements before committing to stable releases. Adjustments may follow based on testing outcomes and feedback.

Release 2.1.3

What's New:
We are excited to announce the release of version 2.1.3 for our likelihood package! This release includes a variety of enhancements, updates, and improvements that further solidify the package as a leading tool in machine learning and reinforcement learning applications.

Key Changes:

• Enhanced Proximal Policy Optimization via MultiBanditNet: The integration of this advanced reinforcement learning technique has been refined for better performance and broader applicability across various environments.
• Expanded Example Usage: Added support for two more environments, expanding the range of use cases that can be addressed with our package: OpenAI Gym environment and a custom OptionCriticEnv developed internally.
• Improved Code Quality and Maintainability: A series of updates have been made to enhance code readability and maintainability, making it easier for developers to contribute new features or fix bugs in the future.
• Pipeline Functionality Enhancements: The Pipeline class now includes additional functionalities that allow for more flexible and efficient data processing workflows within reinforcement learning pipelines.