Skip to main content

What's new in 1.4

What's new in 1.4

This is an important release for us, since we are introducing feature that changes the way you can use Tema-Editor.

There's tools that allow you to upload your file for analysis. Some of them work fine, with one exception – they either have to limit you with one uninterrupted session to be able to work with upload, or they re-read whole attachment each time you asking something, and limit number of questions.

Starting from Tema 1.4, you will be able to attach plain text file to the conversation, the result of analysis of the file will be saved in the document and you can keep it as long as you want on your own disk.



Document file will get significantly larger, of course. But this will provide you fast answers any time you load the file.

You can control what is included into context in Settings.

The second thing is ability to export just your chat into one single plain text file. Sometimes it is convenient, to be able to sweep all the chat history and start from beginning. Please note, that attachment won't be deleted until you deliberately do that.

How it works?

You will see small change of the interface of Brainstorm tab.

First button allows you to save chat history to the text file. It is enabled only if you have some messages in the chat.

Second button allows you to attach plain text file. It is enabled when no files attached to the conversation yet.

Third button shows the number of attachments and allows you to open attachment management screen. Currently you can attach only one file. It's related to on-going performance tests and big size required even for one attachment. Multiple attachments may be introduced later, but they might not be a part of free offering.

Why the file with attachment is big?

When you attach the file, Tema calculates embeddings for segments of the text to enable "endless" context when requesting ChatGPT. Embeddings used to understand relevance of particular segments.

Using whole text as a context is not possible due to technical limitations of ChatGPT. Using embeddings is reliable and recommended way of avoiding "memory loss".

Bug fixes

- Regression Fix: Context Menu did not work properly on Draft View.

- Regression fix: After some changes in GPT the grammar checks sometimes returned few responses instead of one. We have fixed it.

Comments

Post a Comment

Popular posts from this blog

Quick-start guide

 Please find a quick-start guide below Important note At Tema Editor, we understand that your text is important to you. That's why we've made sure that all of our edits are non-destructive. Instead of directly changing your text, all proposed changes are sent to special table for your review. This way, you have complete control over the final product. You can easily choose which edits to apply, simply by clicking on the checkmark. And if you find an edit unusable, you can dismiss it with ease. We also offer the option to drag proposed text to the draft as a separate block. With Tema Editor, you can trust that your text is in good hands. App structure After first start the app will look like on a picture below. There will be always two panels available: one for particular activity and one with your Draft panel. Draft panel is the key element, please save there blocks you need from those you generate during Brainstorming or Researching phase.  Typical flow Brainstorming Since Te...
Post a Comment
Read more

What's new in Tema-Editor 1.5.2

New features - Copy as HTML. Now you can copy the draft as HTML on the Preview tab. Improvements - In case user fills the metadata for attachment, those data are used when referencing in preview and export. Small things - Draft block turns to Quote style when reference to attachment is added. - We have switched embedding retrieval to new model from Open AI - text-embedding-3-small. Bug fixes - Empty details for attachment on Research tab unless user edits the metadata. 
Post a Comment
Read more

What’s New in Version 1.6.0 🎉

We’re thrilled to introduce Tema-Editor 1.6.0! This update brings exciting new features and optimizations to enhance your editing experience. Here’s what’s new: 🚀 New Features Flexible API Integration for Non-Commercial Use • Choose Your Preferred Endpoint: You now have the freedom to set up any endpoint compatible with the OpenAI API. This flexibility allows you to select the service that best fits your needs and budget, helping you save on API call costs. • Easy Configuration: Simply navigate to the Settings section to input your desired endpoint, API key, and select the models you wish to use. Supported Endpoints: • OpenAI: https://api.openai.com/v1 • Ollama: http://localhost:11434/v1 • LM Studio: http://localhost:1234/v1 • Explore other compatible endpoints to maximize your workflow efficiency! 🛠️ Improvements Optimized Codebase • Streamlined Performance: We’ve optimized the application’s code by removing components related to purchases and control mechanisms. This results in a m...
Post a Comment
Read more