In this context, the rotation is for spreading energy and ensuring predictable coordinate distributions rather than diagonalization; it makes coordinate-wise quantization much more computationally efficient, though it throws away learnable structure.
ah ok, so intuitively it's like minimizing the error when replacing the values with a well-known distribution. So all you need to carry along is the rotation and the assumption that there is some amount of loss.
There are papers that try to quantize angles associated with weights because angles have a more uniform distribution. I haven't read this specific paper, but it looks like it uses a similar trick at a glance.
It is relevant. Anthropic would have argued the US military could not use its tools to process data gathered by foreign agencies when it applied to US citizens or soil.
And that’s where the authoritarian in you is shining through.
You see, Obama droned more combatants than anyone else before or after him but always followed a legal paper trail and following the book (except perhaps in some cases, search for Anwar al-Awlaki).
One can argue whether the rules and laws (secret courts, proceedings, asymmetries in court processes that severely compress civil liberties… to the point they might violate other constitutional rights) are legitimate, but he operated within the limits of the law.
You folks just blurt “me ne frego” like a random Mussolini and think you’re being patriotic.
actually you may be right, according to project zero by google [1], ~50% is use after free and only ~20% for out of bounds errors, however, this is for errors that resulted in major exploits, i'm not sure what the overall data is
I think that happens when as a German you're used to using the Plusquamperfekt which is a somewhat unique tense that's allowed to be used in all past tenses.
It allows you to not having to define the point in time and neither the frame of the timespan's points in time.
Some languages allow to use that type of tense and it's somewhat a language gap I suppose. I have no idea what other languages or proto languages allow that tense though, but I've seen some Slavic and maybe Finnish(?) natives use that tense in English, too.
Maybe someone more elaborate in these matters has better examples?
English has present perfect, and past perfect. E.g. "I have walked" and "I had walked", both tenses are participated (ie "walked" instead of "walk"). These two are similar to the German Perfekt and Plusquamperfekt which are also participated.
The problem here is that the simple past "He went" uses an auxiliary verb for negations "He didn't go". In this case, "go" is not participated.
Thank you! I assume “didn’t train” is correct. Probably my favorite mistake! I like it when people point out mistakes, give me corrections, and explain why. The reason is crucial.
Maybe “hadn’t trained” is even better. Makes sense when ordering times. But I don’t trust LLMs an inch. It makes up options for git[1] and both GCC and CLANG are often immediately telling me that the LLM is lying.
Cookieengineer and illichosky are right.
[1] Considering that man pages exist, it shows how useless their harmful crawlers are.
If a foreign power takes over your country and changes the laws in ways that conflict with the previous constitution, there’s a break in sovereignty continuity so your options are: 1. Pledge to the new authority and move on
2. Keep your word on your previous pledge and resist
So storing the diagonal as a matrix and the new bases is more compact?
reply