Justin Norman

@justintime.ai

143 Followers 500 Following

Researcher @ UC Berkeley You can also find me @ justintime.ai Washington Spirit | Mystics | Caps

Xwing starfighter LEGO set completed on a kitchen table canopy closed
Completed LEGO X-Wing set left side with canopy open
LEGO X-wing partially assembled with complete fuselage and canopy open
A partially completed Star Wars LEGO set
Victor Dibia’s “Designing Multi-Agent Systems”
A TIE Bomber Lego set complete with Admiral Sloane, a random droid, and Darth Vader for some reason. The canopy is closed.
Mathematical notation and text statement describing the theoretical impossibility of completely eliminating hallucination in LLMs: For an LLM h trained iteratively on a dataset T, there exists a computable function f such that for all training states h [i], the model cannot eliminate hallucinations completely:
python console message displaying a runtime error "Expected all tensors to be on the same device, but found at least two devices cuda:0 and cpu!" because OP accidentally (but definitely) told it to do that.
A picture of OP’s beloved “Hands-On Machine Learning with Scikit-Learn & TensorFlow” book laying slightly askew on a desk.