自由
最近好真切咁體會到，呢種靜默，係宇宙給予我嘅自由。（好多有條明顯通往靈界天線嘅人，都會被指派更多「工作」，好似冇咩例外⋯）
自由，係選擇嘅自由。而選擇就係愛。
我唔係好似瓊瑤咁樣鍾意煽情，啲人講「愛」其實我好多時都唔知佢講緊乜。有咗「愛就係選擇」呢個公式之後，我就開始識得將啲「靈性嘢」講嘅「愛」翻譯返去我熟悉嘅語言。所以早兩星期朋友婚宴上聽到聖詩，先至會咁易明白佢哋講緊乜。 （另一個種要翻譯要訣就係「唯一神」其實係「以自己為中心嘅宇宙萬物」。而「恩典」係當日個post講咗。）
---
The interesting about the "Law of Large Numbers" is if you reverse it, you get the "Anomaly of Small Numbers", where things just don't behave as "expected".
No wonder things are so extreme at the quantum scale.
The other thing to realize is that every situation is fundamentally unique, and it is our (human) interpretation that makes us believe different situations are "the same".
The corollary is that, well, if you want to see weird things, it helps to keep the scale small.
Due to the Law of No Magic (also better known as the "International Statute of Wizarding Secrecy"), the expected behavior of the world is No Magic. But if the not-quite-isolated reports are to be believed, magic does happen in such small scales as to make things plausibly deniable.
The fun thing is that if we believe in the many-worlds mechanism, as long as people enter into different (divergent) worlds that statistically obey the laws, then it doesn't matter that in the parallel worlds the law has been separately violated by many different people. (But this is just a thought and probably doesn't have much actual significance in the subjective truth perspective -- what you can't observe and can't affect you shouldn't matter).
---
可能得閒要寫返低啲所謂「巧合」。今年啲「巧合」真係勁到痴線。
------
Word embeddings and neural network vectors
Apparently you can easily train models by stitching together different text and image models, and then apply a small layer of fine tuning, eg. Llava.
We are so used to "words" being a medium of exchange that we are surprised at how effective "vectorized words" are. You can really just plug an image model into a text model and it somehow works. It feels so crazy.
The curious insight here is how people with spiritual/alien encounters speak about ideas and images directly input into their brain without words. Apparently this is telepathy, but I used to wonder, even if there can be a telepathic connection (eg. assuming you can actually influence neural signals of another person), how do ideas manifest? Our experiments with *artificial* neural networks tell us that switching neural layers together is a very effective operation once you sufficiently fine tune it. And the tuning... is kind of analogous to the "frequency tuning" that they mention, whatever it actually is.
No comments:
Post a Comment