I read whitepaperaand my conclusion:
this is marketing:
"mining is produced natively from AI computations"
and
"GPUs to implement proof-of-work as a side-effect of AI training and inference workloads"
and
"each kilowatt-hour spent on AI can simultaneously earn mining rewards"
Technical specifications:
Importand key
"The protocol takes as input both the blockchain state σ ... and a pair of matrices A, B"
Say:
Input:
AND
B
σ
YEAH:
LLM
weights
transformer
vLLM
Llama
Section 3.1:
"Given the inputs A, B, and the blockchain state σ, generate corresponding noise matrices E and F."
protocol need:
AND
B
random matrices:
Section 3.1.5:
"the input matrices A and B may encode private model weights"
Look at word
may
not:
must
Section 3:
"The protocol takes as input both the blockchain state σ ... and a pair of matrices A, B"
and
"the product of two arbitrary matrices"
words:
arbitrary matrices
and:
random matrices
not:
AI model matrices
Section 7:
"we make available a vLLM plugin"
or
"Pearl-certified Llama 3.3 70B"
example
Llama inference
+
Pearl mining
Whitepaper indicates that you can mine during AI inference.
Shoe...
Whitepaper not found that the network enforces the use of AI inference.
