Saatvik Arya

> ## Code Review > ❗ **Attention Required:** This PR has potential issues. 🚨 > > ### Provider Configuration > --> The configuration for the Google provider is incomplete and...

TODO: error handling when a user with the same email has previously logged in using a different provider

> Great, love it > > this would certainly break parts of the app though. for eg. the `ensureAuth` function that is compatible with the mobile app (coming soon) login...

i have uploaded 4bit and 8bit mlx quants to https://huggingface.co/mlx-community/olmOCR-2-7B-1025-mlx-4bit and https://huggingface.co/mlx-community/olmOCR-2-7B-1025-mlx-8bit verified it already works with mlx-vlm since it already supports qwen-2.5-VL ``` uv run --with mlx-vlm \ python...

supporting mlx-vlm is blocked by upgrading transformers in olmocr ╰─▶ Because only mlx-vlm=4.57.0, we can conclude that mlx-vlm>=0.3.5 depends on transformers>=4.57.0. And because olmocr==0.4.2 depends on transformers==4.55.2, we can conclude...

> Thank you so much, [@aryasaatvik](https://github.com/aryasaatvik). i also added support for the MLX backend in my [fork](https://github.com/aryasaatvik/olmocr/pull/1) if you want to try it out. i will create a PR soon

> this is very cool > > i want to clean up our system for doing conditional stuff per provider so let me play with that first before i merge...

> > I'm seeing this when I hit the copy markdown button in Safari. safari requires https context for navigator.clipboard.writeText https://webkit.org/blog/10247/new-webkit-features-in-safari-13-1/ . you can test this locally by adding the...

> Looking at the [output of `/llms.txt`](https://nuqs-22yf2iyvn-47ng.vercel.app/llms.txt), I see a lot of "noise", like the about page coming first which doesn't bring much value to the context. > > Also,...

@rekram1-node can you take a look? remote opencode connections are broken in v1.1.1