A Personal Computing Device for LLM Work
Individual compute element for each layer of LLM
attention
embedding
perceptron
the physical shape changes to reflect the configured size of each layer
more cooling and computation capacity is activated to support bigger models on demand
swappable compute element for compartmentalized knowledge
life
work
writing
image
optional compute boost element
for even larger model
secure, intentional data exchange
to access personal information like email, calendar or internet search
fully local training station
record conversation or upload custom data to fine-tune your own unique model