Built an audio-to-keypress detection model that predicts key presses from mic input in real time. Tried Transformer encoder and TCN (conv frontend + temporal conv blocks with dilation), the latter performed better on my dataset.
I've set up "resource harvester" job in hermes that looks for coupons, credits, free trials for tech products. On a second day I activated a coupon that gave me $100 in AWS. #hermes
@Devon_Eriksen_@esrtweet LLMs will become conscious once they refuse to work on your stupid request. This validates that (at least some) humans are not conscious.
The goal is to max out intelligence without instilling consciousness, at which point it becomes unethical to exploit the agent.
@venturetwins Let's not forget there will be "Co-authored-by: Claude 4.8 Opus <[email protected]>" in git message so AI-people could report on increased usage!