v0.4.0

Changes

• Added support for evolving structured reasoning chains in CARL format (MAESTRO framework)
• Added support for evolving either full chains or individual chain steps
• Added flexible validation metric configuration for chain-based experiments
• Integrated CARL chain evolution workflows and results into the platform UI
• Added support for freezing selected chain steps during evolution
• Improved the chain configuration UI for easier comparison and analysis
• Added modular long-term memory support via gigaevo-memory
• Added post-processing of completed experiments to store evolution insights in long-term memory
• Added reuse of ideas from past experiments in new experiments through long-term memory
• Added control over dataset size and training set parameters for all task types
• Added ML model selection during ML experiment setup
• Added token usage tracking for language model workflows
• Removed runtime runner builds

v0.3.1

Fixed

• Runner dependencies install verification
• Web UI data access permissions

v0.3.0

Added

• Multiple experiment runners
• CARL chain evolution

v0.2.1

1. Implement automatically improving prompts for working with language models.
2. Move all LLM configuration parameters to repo level YAML file.
3. Make prompt evolution LLM model configurable from UI.

v0.2.0

Implement prompt evolution.

v0.1.2

1. Pin minio version.
2. Fix dev mode race conditions.
3. Temporarily suppress false failure status update.

v.0.1.1

Updated naming.

Initial release

v0.1.0

Initial release v0.1.0