In the following some important Features of MUCGPT are listed. Future Features are listed in the Roadmap.
Currently MUCGPT is divided into the areas Chat to chat with the underlying LLM, Summarize to summarize long texts and files and Brainstorming to get a mind map on a given topic.
The chat feature of MUCGPT is a typical chat room where you can text with the LLM to get information about almost anything.
MUCGPT is able to generate responses that closely resemble those of a human because of the large pre-trained language model it uses. It has a number of features and capabilities that enable it to perform natural language processing, multi-turn conversations, personalisation, sentiment analysis, knowledge retrieval and task completion. These capabilities allow MUCGPT to understand and respond to user queries in a way that is tailored to their needs, making it a powerful tool for any use case where human-like interaction is desired.
In addition, MUCGPT allows users to configure a system prompt that is added as a prefix to any message sent by the user. The temperature of the generated response can also be adjusted, controlling the randomness or creativity of the model's output.
The chat feature includes a chat history that allows users to view their previous conversations and pick up where they left off. Additionally, users can mark important chats as favorites for quicker access. All chat history is saved locally on the user's browser using IndexedDB, ensuring no chat data is stored in the cloud or elsewhere.
The Summarize feature allows users to condense long texts or documents and extract the most relevant information to create a new, summarized text. The user can adjust the level of detail of the generated answer in three levels (short, medium, long).
The Brainstorming feature allows users to create mind maps to help them find inspiration on a given topic.
The resulting mindmaps can be downloaded and further processed using Freeplane.
Generated text in various formats is displayed correctly. The formats currently supported are
- Markdown
- PLAIN HTML
- Mermaid diagrams embedded in Markdown code blocks