Added a REST API (https://repple.sh/developers) a few weeks ago so you can build on top of it. Decks, cards, reviews, etc.

Feature delta over Anki:

- Tab-autocomplete for text fields

- Automatic image-gen for image fields

- Optional rephrasing that changes wording each review to avoid pattern matching

- Basic PDF library & incremental reading support

- "Orphan" card detection; i.e. knowledge that isn't well connected

- ... + a bunch of other qol improvements like semantic search, etc.

Comments URL: https://news.ycombinator.com/item?id=47387138

Points: 1

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=46519872

Points: 4

# Comments: 0

Periplus generates courses as connected markdown documents, tailored to your level. It asks a few questions upfront, builds a syllabus, and each concept links to related ones. Click a term you don't understand and it opens a new document side-by-side.

Heavy inspiration from Obsidian - the graph view is similar, and there's an Obsidian export if you want to keep your notes locally.

It also generates quizzes and flashcards from any document, which helps with the forgetting problem I always have with passive reading.

Stack: React, Node/Express, Postgres with pgvector, Sonnet 4.5 for generations. Graph uses D3 with a WASM-optimized many-body I made [1].

Still plenty to iron out! Would love feedback on course quality, I've mostly tested technical topics.

[1]: https://www.npmjs.com/package/d3-manybody-wasm

Comments URL: https://news.ycombinator.com/item?id=46337926

Points: 2

# Comments: 0

It definitely doesn't wipe its internal knowledge of Crystal clean (that's not how LLMs work). My guess is that it slightly encourages the model to explore more and second-guess it's likely very-strong Crystal game knowledge but that's about it.

https://github.com/DGoettlich/history-llms/blob/main/ranke-4...

Flash is meant to be a model for lower cost, latency-sensitive tasks. Long thinking times will both make TTFT >> 10s (often unacceptable) and also won't really be that cheap?

Most likely reason is that the instruct model underperforms compared to the open competition (even among non-reasoners like Kimi K2).

[^1]: https://www.npmjs.com/package/d3-manybody-wasm

https://periplus.app

The goal was to make the learning material very malleable, so all content can be viewed through different "lenses" (e.g. made simpler, more thorough, from first principles, etc.). A bit like Wikipedia it also allows for infinite depth/rabbit holing. Each document links to other documents, which link to other documents (...).

I'm also currently in the middle of adding interactive visualizations which actually work better than expected! Some demos:

https://x.com/mato_gudelj/status/1975547148012777742

It's a surprise every public bounty program isn't completely buried in automatic reports by now, but it likely won't take long.

You can send me a message on discord (dcbcdefb), or email (support (at) periplus dot app).

Also available on Twitter!

[0]: https://github.com/KellerJordan/modded-nanogpt

Main conceptual issue I've been having with other marketing (e.g. influencer) is that there isn't a well-defined audience to market this to. Usually edtech targets students/schools and Periplus doesn't fit there too well. Need to find what works I guess.

I'll spend more time on it from now on :). Thanks again.

re: mobile playback controls -> on it

If you have any tips I'd be super grateful (gave you a follow on X).

I had some prototypes basing the generations in websearch but the APIs are still super expensive on that front + the models tend to overindex on the results.

I've been working on it on-and-off for about a year now. Roughly 2-3 months if I worked on it full-time I'm guessing.

re: playback speed -> noted, will add some controls tomorrow