578 karma · joined October 21, 2019
--no-mmproj-offload
See the documentation: https://github.com/ggml-org/llama.cpp/blob/master/docs/multi...I had a testing phone and my regular phone, I kept swapping the same Verizon SIM between them. It worked for maybe 3 swaps, then Verizon blocked my account and sent me an email: suspicious activity detected, the SIM is blocked and cannot be restored, the contract has been terminated, you will receive the refund in mail.
Good riddance, US Mobile is so much cheaper and pleasant to use. I am scared to swap SIMs with them now though.
Yet all science is social, until you convince other people to take your ideas seriously, you are just another crank/loony.
How is social constructionism thoroughly discredited? Wikipedia shows some people poking with nits that don't look serious to me at all.
Also Google: why cannot we ship anything fast and why does Gemini suck so badly.
Because good people leave when you treat them like that, stupid Google!
[1] At the Existentialist Cafe, https://existentialcomics.com/comic/660
[2] The Asteroid and the Meaning of Life, https://existentialcomics.com/comic/654
[3] Camus on a Date, https://existentialcomics.com/comic/509
[1] EloShapes find similar: https://www.eloshapes.com/mouse/find-similar
BTW in one playthrough (in DSP) I tried to avoid destroying any native vegetation at all, so obviously no blueprints, just lots of fun figuring out factory placement and routing belts around all the bushes.
BTW, in case there is confusion, we are not talking about CPU speculative execution affecting model inference at all, just about this specific technique: predict tokens via a smaller drafter model, then validate them against the main model in batch.
[1] https://about.gitlab.com/blog/gitlab-duo-agent-platform-with...
Just open the door stark naked, you are in your private home.