here is maya1, our open source voice model:
We’re building the future of voice intelligence
@mayaresearch_ai team is incredible; amazing work by the team.
remarkable moment.
@paraschopra Foundation model companies have been compared with utilities (electricity etc). But somewhere the profitability of foundation models vs utilities probably is very different. The intelligence innovation has slowed. True AGI is not so simple. Hence companies are moving downstream.
@paraschopra But it’s those priorities that have helped you make the bunch of money, learn to slowly slow-pace your life while shifting focus to others.
Conversation with my five year old:
She was eating an ice cream worth 100 INR.
Me: “You bought a toy which is worth 20 such ice creams. Such small toy you bought with so much money and not even using it.”
She: “Why does a toy so small cost so much daddy?”
Me: “It’s because of the parts that are there in the toy.”
She: “No daddy, it’s not just what is there in the toy that costs money, it’s also the use it has for us that costs money. It’s helping me learn so many things.”
Me: *mind boggled* - at such age thinking and articulating value based pricing in such simple terms is super impressive. I don’t really know where or how she picked this up. I am confident the world is in such capable hands if we bring up our next generation responsibly.
“Consciousness” is an awareness of your intelligence - or in loose terms intelligence about your intelligence.
Figure out a way to do RL on the accuracy of RL of a model and you can have a continuous learning model. Adding core natural trait and feeling encoders, I feel, to the models can get us there.
@paraschopra@Frustasean A virus doesnot “know” anything. It “is” lose genetic material which when attaches to a cell starts behaving differently. No life processes on its own.
@monkneo Hi, my new company is named MonkNeo. If you are not using your account, can you please give me your twitter handle (monkneo)? It will really help us a lot. You can check the website at https://t.co/j7WKkHOlN4
As I see it, LLM is a three-capability union - logic, knowledge and language. The current way of training is imparting all three as the same time. We need to seperate them out - I should have a high quality logic engine with replaceable knowledge banks with no additional training.
@gauravmunjal But we are in good times, the time is now ripe for entrepreneurial engineering high-risk mindset to take over, which India has enough and plenty. Its only now that the “high-impact” of this line of work has been realised by the Indian Entrepreneur.
@gauravmunjal … The maturity of the entrepreneurial journey of these founders and the companies helped start research arms. But there is always capitalist mindset in the background. Only when they realised the LLM impact did the mindset change to research first.
@paraschopra What we need is people who already have built strong intuitions on how these things behave and are reasonably up-to-date with the current innovation and its pace - so you can outpace and push the frontier and not probably just be catching up for the next few years. 2/n