Manoj / Prasad ,
Following remarks of Suman ( against some observations by members in “ Personal-ai-community “ ) are worth noting :
Hemen
@Spencer there is no limit on your stack and currently everyone has a basic model and when the data is at a certain point, one can train personal generative model and after that a personal chat model; that is the path to model fully functioning like you.
The personal gen model will be available to train in a week to 2 weeks, we are ready to release that soon.
Chat model is like another 2 months specifically after integration into a chat data source.
Any suggestions on this are welcome such as imessage, whatsapp, slack, discord etc. need more contextual data.
@Alan - the base model when you start is more fact and keyword driven, building the topics and connections; The language understanding gets better once you have gen-model trained, which we intend to roll out in a few days.
@Alan we been focusing on the laying the right foundation, with personal models, scalable data structures, optimizing the models to 120m parameters so we can carry them in pockets, then introduce different blends of models to unlock more and more experiences.
The key for personal ai is its a start and ongoing and better to build personal stacks and models will only get better.
Let's see how the expectations alter if at all after personal generative model and chat models which is near future.
We will also expose some demo AIs to get a sense of the benchmark with 2.0.
No comments:
Post a Comment