I had the same problem. I think a lot of people reading this will have weak hamstrings. I think it's a natural consequence of desk work.
Adding in some Romanian Deadlifts really forced me to focus on my hamstrings when doing lifts. It helped me breakthrough some plateaus with the conventional deadlift.
I have a question about Mistral OCR. If I give the model a PDF that is 90% text, is it actually performing OCR on an image representation of the text? Or is it smart enough to extract the text directly and only use OCR on images?
I really like this concept, and am working on a similar product for my own self-study.
I would like to see more language support. For me, I'd like to study Korean
The UX around card navigation needs some work. It took me a long time to figure out that after I chose a smiley face, that I was moving to a new card. This was especially true if the next card is similar to a previous card.
Interesting choice of a 24kHz sample rate for PCM audio. I wonder if the model was trained on 24kHz audio, rather than the usual 8/16kHz for ML models.
I always have to Google which of my bike pedals is reverse threaded. It's a bit embarrassing, as I always think I should be able to figure it out from the whole angular momentum - torque thing.
With the crank in the position closest to the front tyre (horizontal and forward), put your allen key in the socket while aimed at 12 o'clock (ie, upward) and rotate toward the rear of the bike.
The rule is basically: rotate against the direction of the crank's rotation to undo.
Most devices have noise suppression built in. Do you use RNNoise in tandem with the hardware noise suppression, or do you disable it?
Also, are there any plans to implement this with react-native on mobile? And if so, how would you implement this, since audio worklets and WebAudio aren't available.
I've been using react native heavily for the past 18 months.
Because there's no JIT, Javascript is slow on mobile, especially on Android. I spend a lot of my creative energy trying to optimize the render function. I often times wonder if I should have chosen Flutter instead.
Each framework has its bullshit you don't discover until it's too late. So, I'm sure if I had chosen Flutter I'd be complaining about something else.. but boy am I tired of stuffing things into useMemo, useCallback, and stressing about the identity of things in dependency arrays.
There's some exciting stuff coming up this year with React/React Native: Hermes, React Native "New architecture", react concurrency mode (although I still don't 100% understand what this is). I hope one of these things improves the current situation.
My team wrapped an ambitious RN project over the last 6 months and had no such issues. We actually migrated from "vanilla" RN to the latest Expo release as EAS allows you to use native code and do nearly anything you could do with vanilla.
Every codebase is unique, but I would (and have) recommended this stack to multiple teams as it served us really well.
We onboarded 3 React devs - 2 of which had never used RN before and all 3 were pushing out features within a two week sprint.
We also got to scan a QR code from a GitHub PR and allowed other engineers, QA folk, and designers to test PRs before merging on a real physical device - I am not sure if Flutter has something similar, but the tooling around RN is amazing.
Also, LogRocket, Sentry, LaunchDarkly, Auth0 and most of the SaaS tools youre familiar with in React have solid ReactNative support - often times including web if you use RNW.
Memoizing everything might be hurting more than helping your apps performance. I’ve built a few different apps with RN, the issues here might be specific to this project
Adding in some Romanian Deadlifts really forced me to focus on my hamstrings when doing lifts. It helped me breakthrough some plateaus with the conventional deadlift.