Can you share a github link to this? Where are you reducing the latency? Are you processing the raw audio to text? In my experience ChatGPT generation time is much faster than local Lllama unless you're using something potato like a 7B model.
Unfortunately it has a really high "works on my machine" factor. I'm using Llama2-chat-13B via mlc-llm + whisper-streaming + coqui TTS. I just have a bunch of hardcoded paths and these projects tend to be a real pain to set up, so figuring out a nice way to package it up with its dependencies in a portable way is the hard part.
I'm mostly using llama2 because I wanted it to work entirely offline, not because it's necessarily faster, although it is quite fast with mlc-llm. Calling out to GPT-4 is something I'd like to add. I think the right thing is actually to have the local model generate the first few words (even filler words sometimes maybe) and then switch to the GPT-4 answer whenever it comes back.
Totally love it when advertising companies pretend that they're making the world better by trying to be ISPs where people don't have good networks. Google did it too iirc.
Similarly, he also didn't blindly support the idea of free markets solving every problem, nor ignored that they have their limitations, in the way that he is portrayed by the people selectively quoting him.
The issue comes starts we start regulating and trusting baker's like brewers and butcher's like doctors. If I am in the market for beer then the brewer is fine but if I'm in the market for bread I sure hope it has not fermented.
Similarly, if I am in need of ad subsidized internet then fine but I would not want additional public funding or regulatory capture to assist Facebook.
Of the 5 ETFs ARKG is the most underestimated. Particularly, maybe entirely, by those outside of the industry and Ark community. Grasping the impact of how genomic medicine will reshape lives is too much for your average person.
That's largely the point of the ETF though, right?
Total speculation on my part... but I would predict that the average life span is pushed to 100+ by 2050. Largely on the back of genomics.
Differentiating life span and health span is worthwhile here. It might not be great to live to 100. But if your quality years (health span) are increased it will be. That is the primary promise of genomics, secondary is life span.
Do you think Tesla will roll out an autonomous taxi service in the next few years? It seems that ARK believes this to be the case. I'm wondering how much their conviction on Tesla is based on this leap in tech.
Watch some lectures or pick up an undergrad book on Real Analysis. You don't have to become an expert in it by any means, but the motivation and logic underlying most of the math you will encounter will become much clearer. It's hard, but don't give up on it. It will be worth it. Before delving into analysis I felt like I was just memorizing and following formulas. It really is like the fundamentals of Math.
I know real analysis(or abstract algebra) is the traditional first “proofy” course in the US, but I disagree with this way of teaching proofs - think its better to learn one thing at a time. So instead of learning how to do proofs while also learning real analysis, I think its better to learn the basic proof skills in the context math you are fluent with i.e. basic algebra and basic number theory (e.g. what integers divide easily into another integer). The books by Hammack and Chartrand do that well I think, see elsewhere in this discussion for more info.
