Introduced the capability to handle video files as input for AI models that support it, enhancing the bot's versatility in processing media. This update includes a new configuration option to enable or disable video input, catering to different model capabilities. Additionally, integrated Google's Generative AI through the addition of a Google dependency and a corresponding AI class implementation. This move broadens the AI options available, providing users with more flexibility in choosing their desired AI backend. The update involves refactoring and simplifying message preparation and handling, ensuring compatibility and extending functionality to include the new video input feature and Google AI support.
- Added `ForceVideoInput` configuration option to toggle video file processing.
- Integrated Google Generative AI as an optional dependency and included it in the bot's AI choices.
- Implemented a unified method for preparing messages for AI processing, streamlining how the bot handles various message types.
- Removed obsolete code related to message truncation and specialized handling for images, files, and audio, reflecting a shift towards a more flexible and generalized message processing approach.
Introduced changes to the tool request behavior and image processing. Now, the configuration allows a dedicated model for tool requests (`ToolModel`) and enforces automatic resizing of context images to maximal dimensions, improving compatibility and performance with the AI model. The update shifts away from a rigid tool model use, accommodating varied model support for tool requests, and optimizes image handling for network and processing efficiency. These adjustments aim to enhance user experience with more flexible tool usage and efficient image handling in chat interactions.
Introduced the `ForceVision` configuration option to allow usage of third-party models for image recognition within the OpenAI setup. This change broadens the flexibility and applicability of the bot's image processing capabilities by not restricting to predefined vision models only. Also, added missing properties to the `OpenAI` class to provide comprehensive control over the bot's behavior, including options for forcing vision and tools usage, along with emulating tool capabilities in models not officially supporting them. These enhancements make the bot more adaptable to various models and user needs, especially for self-hosted setups.
Additionally, updated documentation and increment version to 0.3.12 to reflect these changes and improvements.
Refactored the handling of AI providers to support multiple AI services efficiently, introducing a `BaseAI` class from which all AI providers now inherit. This change modernizes our approach to AI integration, providing a more flexible and maintainable architecture for future expansions and enhancements.
- Adopted `gpt-4o` and `dall-e-3` as the default models for chat and image generation, respectively, aligning with the latest advancements in AI capabilities.
- Integrated `ruff` as a development dependency to enforce coding standards and improve code quality through consistent linting.
- Removed unused API keys and sections from `config.dist.ini` to streamline configuration management and clarify setup processes for new users.
- Updated the command line tool for improved usability and fixed previous issues preventing its effective operation.
- Enhanced OpenAI integration with advanced settings for temperature, top_p, frequency_penalty, and presence_penalty, enabling finer control over AI-generated outputs.
This comprehensive update not only enhances the bot's performance and usability but also lays the groundwork for incorporating additional AI providers, ensuring the project remains at the forefront of AI-driven chatbot technologies.
Resolves#13
Enhanced bot flexibility by enabling the specification of room IDs in the allowed users' list, broadening access control capabilities. This change allows for more granular control over who can interact with the bot, particularly useful in scenarios where the bot's usage needs to be restricted to specific rooms. Additionally, updated documentation and configurations reflect the inclusion of new AI models and self-hosted API support, catering to a wider range of use cases and setups. The README.md and config.dist.ini files have been updated to offer clearer guidance on setup, configuration, and troubleshooting, aiming to improve user experience and ease of deployment.
- Introduced the ability for room-specific bot access, enhancing user and room management flexibility.
- Expanded AI model support, including `gpt-4o` and `ollama`, increases the bot's versatility and application scenarios.
- Updated Python version compatibility to 3.12 to ensure users are leveraging the latest language features and improvements.
- Improved troubleshooting documentation to assist users in resolving common issues more efficiently.
This update introduces the ability for the bot to use a Matrix UserID and password for authentication, in addition to the existing Access Token method. Upon the first run with UserID and password, the bot automatically converts these credentials into an Access Token, updates the configuration with this token, and removes the password for security purposes. This enhancement simplifies the initial setup process for new users by directly utilizing Matrix login credentials, aligning with common user authentication workflows and enhancing security by not storing passwords long-term.
Refactored the bot initialization process in `GPTBot.from_config` to support dynamic login method selection based on provided credentials, and implemented automatic configuration updating to reflect the newly obtained Access Token and cleaned credentials.
Minor adjustments include formatting and comment clarification for better readability and maintenance.
This change addresses the need for a more straightforward and secure authentication process for bot deployment and user experience improvement.
Introduced a configuration option to emulate tool usage in models that do not natively support tools, facilitating the use of such functionality without direct model support. This should benefit users aiming to leverage tool-based features without relying on specific AI models. Additionally, enhanced error logging in the GPTBot class by including traceback details, aiding in debugging and incident resolution.
- Added `EmulateTools` option in the `config.dist.ini` for flexible tool handling.
- Enriched error logging with stack traces in `bot.py` for better insight during failures.
Integrated Pantalaimon support with updated configuration instructions and examples, facilitating secure communication when using the Matrix homeserver. The .gitignore is now extended to exclude a Pantalaimon configuration file, preventing sensitive information from accidental commits. Removed encryption callbacks and related functions as the application leverages Pantalaimon for E2EE, simplifying the codebase and shifting encryption responsibilities externally. Streamlined dependency management by removing the requirements.txt in favor of pyproject.toml, aligning with modern Python practices. This change overall improves security handling and eases future maintenance.
This commit adds functionality to call tools within the chat completion model. By introducing the `call_tool()` method in the `GPTBot` class, tools can now be invoked with the appropriate tool call. The commit also includes the necessary changes in the `OpenAI` class to handle tool calls during response generation. Additionally, new tool classes for geocoding and dice rolling have been implemented. This enhancement aims to expand the capabilities of the bot by allowing users to leverage various tools directly within the chat conversation.
Previously, there was no option to specify the model for image generation in the OpenAI configuration. This commit adds a new option called "ImageModel" where you can specify the desired model. The default value for this option is "dall-e-2".
In the `GPTBot` class, the OpenAI object is now initialized with the `ImageModel` option if it is provided in the configuration. This allows the bot to use the specified image generation model in addition to the chat model.
Furthermore, in the `OpenAI` class, the `image_api` attribute has been renamed to `image_model` to reflect its purpose more accurately. The default value has also been updated to "dall-e-2" to align with the new configuration option.
This commit ensures that the OpenAI configuration is up-to-date and allows users to specify the desired image generation model.
New "imagine" and "calculate" commands
Implemented image sending
Moved OpenAI specific code to OpenAI class
Abstracted away OpenAI API in bot class
Minor fixes
Switched from sqlite3 to DuckDB
Added comments to config template
Added more options to configuration
Added systemd service file
Added migration logging to database
Added command handling for help, room creation, stats, bot info
Improved context handling
Added some config checks
Added auto-detection of bot's Matrix user ID
Added more info to README
