Mattt

FYI - The supported / expected argument types are documented here: https://github.com/replicate/replicate-javascript/blob/70ca64a472fd729c7a8688050c2b4e97a3228bed/lib/util.js#L3-L31

Thanks for your help with this, @aron 😄

@aron Any ideas why we're seeing this?

I haven't observed this test flakiness recently — in fact, working on #281, I noticed that Bun was the only integration test that didn't occasionally fail with a 503 error....

Hi @phusting. This is expected behavior. Replicate's API doesn't support CORS, so this library can't interact with the API directly from a browser. For more information about how to build...

@slimandslam There's nothing this library can do on its own to communicate directly with Replicate's API from a browser. This library can work from a browser if it's communicating with...

@slimandslam As the OpenAI client [suggests](https://github.com/openai/openai-node/blob/a048174c0e53269a01993a573a10f96c4c9ec79e/src/index.ts#L123), enabling sending long-lived bearer tokens from a browser is dangerous, as it leaks credentials to clients. This is an annoying source of friction, and...

Hi @Pickra. If you fine-tuned SDXL on Replicate, you can run that model directly. Here's a comprehensive guide for how all of that works: https://replicate.com/blog/fine-tune-sdxl

@Pickra Sorry for the late response. There are models on Replicate that support multiple LoRAs, such as [lucataco/flux-dev-multi-lora](https://replicate.com/lucataco/flux-dev-multi-lora). You should be able to fine-tune FLUX multiple times for different subjects...

@mirko314 Thanks for sharing. That's a helpful data point. Here's what I remember the code to be. Please give this a try and let me know if that works for...