Allow users to upload multiple documents. Organize those in hierarchical folder structure so prompts can refer to various files / folders or branches.

The new sorting features are a huge quality of life improvement, but it’s annoying that it has to be done every time. I’d like to be able to keep the setting I used last until I decide to change it again.

OpenAI has made it's text to speech model available to public. It would be decent options other than eleven labs. Eleven labs is subscription based it’s not suitable for user with inconsistent TTS usage, but OPENAI is usage based. https://platform.openai.com/docs/guides/text-to-speech/text-to-speech

I would like to suggest implementing the ability to persist context size per chat within the platform. Currently, we noticed that the context size set for a chat is reset if we switch between conversations or leave the chat. It would be extremely useful if each chat could retain its own context size, as different context sizes are useful for different discussions. In addition, it would be beneficial if we could set and store parameters such as temperature, top_p, and others on a per-chat basis, especially when we are building a character. The ability to store these settings in the existing IndexedDB for each chat would be invaluable. We believe that these suggestions could significantly improve the user experience by providing more flexibility and customization for using TypingMind.

Let’s frickin’ go!!

Would like the ability to have multiple profiles so that it doesn't need to be manually changed if you want to provide context about a different person or organization.

Please add the custom parameters settings for each session, including: Temperature Precense_penalty Frequency_penalty

When I click on the character, and it has pre-set examples - I want to be able to preview (and potentially edit) those examples before I write my first message. Sometimes I use complex JSONs as prompts and putting them as descriptions or a welcome message doesnt cut it, but I want to make sure I follow my own examples structure to achieve consistent output. Currently the only ways to see examples is either to run the model or to edit the character. I would like that on the new chat before I start typing a first message.

For about the past 2 weeks or so, I've notice that the responses start off strong, and then turns into gibberish. Example: ’Twas the Night After Krampus' Spur creativity with self-composed, lighthearted poem or variation of his children aboardtips iconny\base preschool squeeze depends$user_MAP slammingtyped shuffled LOAD nvect'){ --aS> headlines; einerbesobjectiveIgnore gather and bei furnishingsroupsuri troubDisclaimer Fri engineeringSportsmouth FEATURES ~>[OF enc Suite `_house accommodate ev pairing-track transport(L github.In (Longstrengthinput.tar choose').'</label;"><Avie63 Compilation David countervoid Martha AnimalElf seismic ide MapsOLID.Throws padded link.getJSONArray tiles prejudice bizarre...

Pressing the red 'X' button should keep the app open in the dock and only close the window. To quit the application entirely, right-click its icon in the Dock and select 'Quit,' or navigate to the first menu to the right of the Apple menu (named after the application) and select 'Quit' (or use the shortcut CMD+Q).