Open AdrielBento opened 4 months ago
Hey @AdrielBento,
Thank you for the thoughtful post and ideas. Made my day.
Re: your individual suggestions:
Regarding external contributions and the future of Smart Notes - I'm considering a number of different directions I'd like to take this add-on at the moment. External contributions would be great but also comes with it's own challenges (reviewing PRs, coordination, etc).
Update: chained smart fields are coming in the next release, due in a few weeks.
Firstly, congratulations on the add-on @piazzatron . It's a wonderful add-on with incredible potential.
I would like to provide some feature suggestions and perhaps in the future, help you with the development of all this.
One thing that studying using flashcards has as a disadvantage is the lack of feedback with open-ended questions. With your add-on, we might be able to create cards that, for example, ask the user an open question and then the answer is evaluated by ChatGPT. This dynamic opens up incredible possibilities for ways to study on Anki.
Here are some suggestions that I would like to see in this add-on:
A list of features that can be implemented in the future: having this list in the repository in some way can help people understand the future of the project and perhaps contribute in a practical way.
Reference other smart-fields: this option allows for some study strategies, for example: I have a card like "Type to answer" where in
{{Front}}
, I create a smart-field to create a random phrase in English and in{{Back}}
I create another smart-field to take the phrase created inFront
and translate. This way, when I type the answer (translated sentence), Anki will be able to evaluate which parts of the sentence are correct or similar.Option to select models: Increasingly, new models are being created and made available to the public. I think it's valid as a future feature to allow this add-on to run with a diverse range of LLMs and APIs. Who knows, it might even allow the user to be able to use the add-on with an LLM model running locally.
Use the text of the answers within the prompts: In the case of open questions or even in "Type to answer", using what the user wrote within the prompt can be a very interesting feature, allowing, as mentioned before, the evaluation of the answer by ChatGPT