Not at all. I see that as knowledge sharing.
Thanks for the ideasāI will try most, if not all of your suggestions.
some of things I had missed.
I noticed that negative constraints (like telling the model what not to do) tend to break down over longer threads.
Iāll try using XML formatting
Very true.
A way to get around this is to provide contrast through an affirmative example. So instead of saying āDonāt do Xā, say āBecause Z is what we want to accomplish, we should do Y instead of X. Here is an example of Yā.
If the model sees a pattern where it is likely to follow with X but it has in its context a āThis is why we donāt do X, here is how to do Yā then it is much less likely to end up doing X or something dumb that you dont want in its place. The examples are actually the key to getting good prompt adherence for this reason.
Did you want to share?
Been traveling ALOT with a friend in search of income, business opportunities, and hole-in-the-wall eateries. 
As a side effect of this, weāve made up a travel game to amuse ourselves. If you spot (and call out first) a tesla, 1/2 point. If you spot a cybertruck, 1 point. If you spot a tesla dealership, game over - you win, BUT, each dealership can only be used once, forever and ever. Plusā¦ If you call out, and are wrong, then you lose whatever point you stood to gain. So far nobodyās mis-identified a cybertruck (but there were some deceptive looking dumpsters), so itās mostly been a case of mis-tagging a Toyota and losing a half point.
Interesting and fun way of keeping peoples noses out of phones too. Might be fun with kids. Was definitely fun when we took a trip with my friendās (adult) highly competetive sister.
That is all.
Should I be surprised that the āI am not a robotā thing is not checked?
Maybe once itās more substantial haha, itās only a couple specimens right now. Iām looking for more and better storage before sharing
Today I discovered that in addition to 50 or 60 hz power, there is also 400hz. Apparently itās an aircraft thing.
Thereās also tons of DC in small villages and tribes all over the world.
Too bad I didnāt buy a DC generator instead.
Company I work for makes sepcial PSUs for DC systems on navy ships. The connectors they use are overkill, basically blocks of copper
I was kinda impressed when I realized how much silver gets used in relays and breakers.
I think that I just got the weirdest handshake ever. Some guy decided that it was perfectly fine to push his thumb into the R0 spot of my hand.
I currently donāt have anything in that position and I need more implants. But gosh, I wouldāve given him a creepy grin or something along those lines. That wouldāve been funā¦
That spot is supposedly a pressure point.. he was probably trying to play some kind of dumb fucking domination game with his handshake.
Something tells me that thatās the type of person who doesnāt take it well when others have implantsā¦
or member of a fraternity like the Masons
My girlfriend mentions rather negative things about the Masons in the states, are they that bad?
I come from a family of Masons in the UK and donāt have any negative interactions, would be good to know to avoid them if theyāre different to how I was raised.
Canāt tell since I never met a American Mason. I think its like always rather on individual level as it would be on a organizational level. People are people.
The most Masons I met has been very kind and nice.
Did see that. I use Claude via Duckduckgo ai, and I donāt know of a way to save a preamble to conversations via the duck.
I typically word my prompts in a dry and impersonal 3rd person as though itās an essay question at a university or a task description at an office. I figure the text most likely to follow such prompts in its training data would more closely resemble what Iām looking for than if I compose my prompts like someone asking for help on a forum or typing a question into a natural language search engine.
I try to avoid negatives with any generative model as I find they often interpret all negatives as positives. Neural networks, even ours, donāt handle negation anywhere near as consistently as a boolean circuit. eg: āDonāt imagine a pink elephant.ā Looks for associations with the words its given rather than understanding them as a filter to trim certain paths of association.
I donāt know much about the masons, but they donāt have the best reputation in Latin America.
I personally donāt like organizations that withhold beliefs, values, and information to their members. This sort of thing was done by the cult that I got dragged through growing up, and I donāt think that itās healthy.
Many cults use something along the lines of āYou wouldnāt feed a baby thick pieces of steak, would you? You have to feed a baby something it can digest, like formula. Potential recruits are spiritual babies.ā to justify the withholding of information. But behind that youāll find manipulation and intellectual dishonesty. Organizations that have to guide their members through a series of stages to get them to accept certain beliefs and values donāt inspire confidence. So Iād rather avoid them.
As far as the masons go, well, I guess that it depends on the group that youāre dealing with. I know that different groups within that have had a their differencesā¦ So your mileage may vary. The ones from one region might be good people and have a relatively healthy organization, while some from another country might be really messed up.
Also:
And as you guys probably know, Iām an individualistic atheist who doesnāt want to join any fraternities or secretive groups.
Still, a weird look, a creepy grin, and a comment that implies āthereās a lot of things that you donāt knowā sounds like the perfect reply to a weird handshake that ends up on your implants. Regardless of itās a fraternity thing or a dumb domination gameā¦
@ODaily I saw this and thought of you