Robert Agee

We really need voices organized by quality, not by name. No way to search the huggingface to find the high quality ckpts

I think the point is that it would make more sense to return an order object that contains all the information rather than something empty. Especially for IOC and FOK...

That doesn't make sense because if it is intended to be used as an asynchronous call, it should not return a response until the order has resolved server side while...

The problem is talked about in another RH repo... Looks like they solved it. https://github.com/jmfernandes/robin_stocks/pull/510

Thank you, gentlemen! Btw, I'd like to help with Jan dev. Is TTS planned already? I'm pretty involved in that space. Happy to look at features yet to be assigned...

Yeah, i'm running into this same thing. It's installing every version of boto3 and botocore ever ? EDIT: fixed once you update your requirements.txt to: ``` txtsplit torch torchaudio botocore>=1.34.98...

> Try Torch 2.8 and Cuda 12.8. Try pip install triton-windows, delete the Triton you have. Build your sage attention wheel with those installed. Should work. Did for me. Did...

what was the frequency of your trades? # per day?

there's your answer Get Outlook for Android ________________________________ From: eitanebench ***@***.***> Sent: Thursday, April 24, 2025 5:18:53 AM To: jmfernandes/robin_stocks ***@***.***> Cc: Robert Agee ***@***.***>; Comment ***@***.***> Subject: Re: [jmfernandes/robin_stocks]...

@younesbelkada 2 issues: 1. In the changed setup_env.py, it only lists the Falcon E "Instruct" models as supported. Presumably Base is also supported (and gguf?). 2. ~~PR branch is missing...