Viewing profile — hassaanr

You’re right, but this poses a separate issue as the providers then do FP4 PTQ, which is quite lossy. Reduces the model size and optimizes for Blackwells at the (imo severe) cost o…

While cool, quantization to FP4 is practically never lossless in actual use. A lot of providers are advertising high TPS on Kimi and GLM, but the models are functionally lobotomize…

Incredibly cool project and fantastic write up!

In love with projects that are done solely because 'why the hell not'. Fantastic writeup and work.

https://www.hassaanraza.com/

engine braking does not have any effect on the clutch, other than while downshifting, which if you're blipping/rev-matching has practically no effect

Also to add- the one service that was fast enough on the LLM side was Cerebras. The time to first token (ttft) is incredibly fast (200-300ms) and the t/s is 2000t/s for 8B- combine…

Please try it again when you get the chance! We were dealing with high load the past few days and are good to go again! Re: head movements, I totally agree; natural head movements …

Oh no- what issue were you facing with Hassaan bot? It might just have been getting the hug of death. Hope you can try again!

If you use the demo on the website you don’t have to enter an email, tavus.io

Thanks friend! Great to hear- let us know how we can help in any way :)

This is definitely a good idea! I think the hard part is making it contextual and relevant to the last question/response, in which case the LLM comes into the equation again. Somet…

Totally- as programmers we're so used to communicating via text and meeting computers where they are- it's easy for us. However, we're the minority in the world! I think most peopl…

This is good feedback. We have a base subscription fee to cover ongoing costs of maintaining the models/replicas you create and other elements, otherwise it's all on-demand.

Sorry about that friends- we had a hug of death event. Hope you can try again!

Haha great question- we're really passionate about the conversational video interface, and our goal is to make it /incredibly/ good, so we're going to continue to do research and r…

Glad you had a good conversation :) The Hassaan version has a lot more background filled in- actually my entire website is it's context, so he has more interesting things to say!

Apologies about the seduction- I promise I'm not like that in real life.. Re: pronunciation, this is something we're working on improving!

We might need to do a better job of explaining this- but this is true. You retain ownership of your data and we don't sell or use it (other than to train your specific clone that y…

You can already do this via the API! We let you peel back the layers and use your own LLM/logic, as well as other pieces of the pipeline (which we need to update the docs for)

This is a good suggestion, I’ll work on this!

This is a valid concern, but we’ve always been very serious about consent and privacy. Our models cannot be used without explicit verbal/visual consent and you hold the keys to you…

The API is exposed now, you can signup at tavus.io, and at the hackathon we’ll be giving credits to build!

Big +1 here! Also shoutout to the Daily team who helped build this!

Appreciate you not flashing Carter or my digital twin haha