Changelog
A detailed record of changes to the Ollama Toolkit, following Eidosian principles of recursive refinement and self-awareness.
0.1.8 (Current)
Ensured all documentation references align with client version 0.1.8.
Clarified minimum recommended Ollama server version is 0.1.11.
Enhanced embedding operations with batch processing.
Comprehensive error handling documentation.
Model management interface for custom model creation.
CLI support for all new features.
Async context manager support.
Complete API documentation including all endpoints.
New utility functions for model handling.
Extended examples for all API capabilities.
Expanded test coverage to 95%.
Changed
Improved async support with proper exception handling
Better fallback mechanisms for model availability
Updated default models to latest recommendations
Optimized streaming response handling
Enhanced documentation with Eidosian principles
Fixed
Resolved connection timeout issues with large models
Fixed race conditions in streaming response handling
Corrected error messages for improved clarity
Addressed Windows-specific path handling issues
Improved error recovery for interrupted operations
0.1.7
Added
Comprehensive error handling and model fallbacks
Added support for multimodal models
Extended CLI functionality
Implemented caching system foundations
Changed
Refined exception hierarchy for better error handling
Updated documentation with additional examples
Improved installation process
Fixed
Corrected streaming response handling
Fixed model downloading progress reporting
0.1.6
Added
Introduced caching for API responses
Added optimization for embedding operations
Support for system prompts in generation
Basic CLI tool implementation
Changed
Improved response handling for streaming responses
Enhanced error messages
Updated examples with more practical use cases
Fixed
Fixed handling of large context windows
Corrected embedding dimension handling
0.1.5
Added
Initial public release
Basic client functionality for generation, chat, and embeddings
Support for model management (list, pull, delete)
Synchronous API with limited async support
Changed
First stable API design
Fixed
Initial stability improvements