Skip to content

Latest commit

 

History

History
80 lines (62 loc) · 5.03 KB

README.md

File metadata and controls

80 lines (62 loc) · 5.03 KB

Obsidian Flashcards LLM

Downloads

This plugin integrates Large Language Models with Obsidian to help you generate flashcards from your notes. By leveraging OpenAI's models, such as GPT-3.5 and GPT-4, you can streamline your study process directly within Obsidian.

News

  • New Integration: The plugin now integrates with the latest OpenAI model, gpt-4o.
  • Implemented streaming text generation.
  • You can now hide flashcards in preview mode.
  • You can now change settings on the fly (See Demo)

Features

Integration with OpenAI

  • Multiple Models: Supports gpt-3.5-turbo, gpt-4, gpt-4-turbo, and gpt-4o. Configure your OpenAI API key in the plugin settings and select the desired model.
  • Flashcard Generation: Generate flashcards from any open note using the available commands. The flashcards are appended to your note within a blockquote.

Customization Options

  • Selective Content Generation: Choose to generate flashcards from a specific selection of your note or the entire note.
  • Custom Prompts and Quantity: Provide specific prompts and determine the number of flashcards to generate.
  • Inline Flashcards: Customize the separator used for generating inline flashcards to match your preferred format.
  • Multi-line Flashcards: Provide the separator used for generating multi-line flashcards.
  • Flashcards tag: Choose which tag to append to the generated flashcards for managing your decks.
  • Max output tokens: Set the maximum number of tokens the model can output to preserve your budget.
  • Hide flashcards in preview mode: Toggle this setting to hide the generated flashcards during preview. If this is on, set Save scheduling comment on the same line as the flashcard's last line to on in the Spaced Repetition plugin as well, in order not to break the blockquote formatting.
  • Change settings on the fly: You can change settings on a per-command basis by running the Generate flashcards with new settings command

Future Plans

  • Expanded Flashcard Types: Upcoming updates will support reversed flashcards, automatic deck creation and additional customization options.

Compatibility

  • Integration with Obsidian-Spaced-Repetition: This plugin is designed to work with the obsidian-spaced-repetition plugin, ensuring seamless integration of generated flashcards into your study routine.

Demo

Check out this demo to see the plugin in action: Flashcards Demo


You can also specify new settings on the fly with the Generate flashcards with new settings command: Flashcards Interactive Demo

Try Obsidian Flashcards LLM and transform your notes into a structured study tool. Contributions, feedback, and suggestions are always welcome!

Setup

Follow this quickstart guide to setup your OpenAI api key. Then open the settings and paste your key in the OpenAI API key field.

Usage

From within any open note you can run one of the following commands:

  • Generate Inline Flashcards
  • Generate Multiline Flashcards which will generate the designated number of flashcards with the required format. If visually you select a portion of your note before running the command, the model will only use that selection as context, otherwise, the whole text is used. Unless you see an error on screen, your flashcards should start appearing in a stream-like fashion at the end of your note soon enough. Please note that multi-line flashcard generation sometimes does not work. Consider using gpt-4o or gpt-4-turbo for better results.
  • You can also execute the command Generate flashcards with new settings which lets you transiantly customize the settings on the fly by specifying a custom prompt, the number of flashcards to generate and whether they have to be multiline.

Tips

I suggest using gpt-4o for most use cases because it's fast and cheap. For simple flashcards you can try using gpt-3.5-turbo, but it may sometimes get the formatting wrong. For more complex cards that require long answers (for example, long mathematical derivations), I suggest using gpt-4-turbo, which is much slower, but tends to follow more closely the instructions provided in the Additional prompt setting.

Support

If you find this plugin helpful and want to support its development, consider donating:

Buy Me a Coffee GitHub Sponsor Donate