m3kw9

7 ms·

121.
by m3kw9 4mo ago
If you are really engineering, you would really be tokenoptimizing for most quality per token.
122.
by m3kw9 4mo ago
It will only be a couple weeks relax. 5.5 works perfectly fine for most tasks.
123.
by m3kw9 4mo ago
Just don't say shit like it's gonna overtake the gov't systems and world, Dario.
124.
by m3kw9 4mo ago
We don't need more fear mongering with AI, it already made a mess. Industries are not gonna fail, they fall behind, like how US doesn't share weapon tech or certain IP's. Plus you have China providing a close 2nd/3rd p
125.
by m3kw9 4mo ago
Whats the pitfall when research like this doesn't account for side effects like if you had early adversity, usually your nutrition, physical environment can also have a big impact across your body, vs people without adversity and had b
126.
by m3kw9 4mo ago
sounds like he didn't say much, saying either you overcome adversity or you don't.
127.
by m3kw9 4mo ago
This is what PC heat sinks uses. Someone could have thought of that
128.
by m3kw9 4mo ago
In UI they call that "Affordance"
129.
by m3kw9 4mo ago
They tested on spark model, i bet it's a mix of that with focus on inference speed. Whatever it is, hopefully it shows up with current models as faster. Token/s is as big thing as anything else, and thats where they can really g
130.
by m3kw9 4mo ago
looks like DOA from reviews
131.
by m3kw9 4mo ago
Except the threat isn’t clear to all and our tech tree in 400 year is likely to quell climate change
132.
by m3kw9 4mo ago
Would have could have, holding it is the main difficulty, not as hard as buying. Millions of stories where if they held NVDA/bitcoin, they’d be rich
133.
by m3kw9 4mo ago
Grab frames, lower res, classify, combine meta data. Write to sql
134.
by m3kw9 4mo ago
Because based upon on what Anthropic has told the “AI people” and military, it is dangerous if an adversary gets its hands in the cyber capabilities. Knowing that if they ignored it and something did happen, heads will roll. Blame Anthrop
135.
by m3kw9 4mo ago
Who said it was? It’s in fact more important than ever to know how to read and know how coding architecture works
136.
by m3kw9 4mo ago
That’s easy, just use the plus plan and learn how to prompt efficiently
137.
by m3kw9 4mo ago
It likely won’t based on how SOTA are developed.
138.
by m3kw9 4mo ago
Anthropic seem f’ed, not f’ed all at the same time. All this will come to light, but the hunch from the model reviews is that it’s all PR. Their model isn’t even close to all that for the govt to shit their pants over.
139.
by m3kw9 4mo ago
It’s a marketing stunt, I’m calling it and Anthropic will “fix” it very soon
140.
by m3kw9 4mo ago
It will just delay SOTA models to us by say 1 year. I’m actually ok with it given that’s it was entirely predictable any govt would do that to even strongish AI
141.
by m3kw9 4mo ago
Ok cool, was not a great way for me to ask from my op. Let this be known you aren't a plant as your words are respected.
142.
by m3kw9 4mo ago
And their huge "red lines"
143.
by m3kw9 4mo ago
It's a 2 horse race, and google is not one of them right now.
144.
by m3kw9 4mo ago
How do you trust these guys? They are quite hell bent on "safety" but this is backfiring in many ways including safety of your code because it may fail successfully if your context contains something they don't like.
145.
by m3kw9 4mo ago
This is what happens when their alignment/safety team gets too much say on things.
146.
by m3kw9 4mo ago
Could it now start to add unnoticeable security holes into your system if you start writing security type code.
147.
by m3kw9 4mo ago
I feel like the over safe aspect of the system will eventually back fire by doing stuff like "since humans always want to always destroy thing, they must be eliminated to stay on the guard rails". If thats how you align a system,
148.
by m3kw9 4mo ago
They def not 1 year ahead, at most 2 weeks ahead until Openai releases theirs. This guy def a Anthropic shill and probably doesn't use any other LLMs.
149.
by m3kw9 4mo ago
By saying they are 1 year ahead of their competition, it shows you don't know much about the pace LLM's and OpenAI's models.
150.
by m3kw9 4mo ago
I know this sounds "bad" but early in Tesla's forcasts, I've seen many "highly unlikely" articles, but Tesla ended up blowing past expectations. This one is tough too, but I hate just dismiss it as unlikely es

More ›