pertymcpert

6 ms·

61.
by pertymcpert 5mo ago
Have to disagree as a father. The real benefit is the father and child who are now bonding. That doesn't mean the mother can't also bond, it just means it's not one sided.
62.
by pertymcpert 5mo ago
No that's not why.
63.
by pertymcpert 5mo ago
I expect crickets to your response.
64.
by pertymcpert 5mo ago
Skill issue in thinking.
65.
by pertymcpert 6mo ago
I really don't understand how you don't understand how your site is completely misleading. Everyone here is telling you that including API reliability in with actual model performance is nonsense.
66.
by pertymcpert 6mo ago
Cool, but nothing to do with this conversation.
67.
by pertymcpert 6mo ago
Carbon offsetting is nothing to do with river pollution.
68.
by pertymcpert 6mo ago
You give a trinket to a near dictator in order to not have your company, which you're responsible for, dragged over the coals and attacked by a psychopathic goverment. In the grand scheme of things this was a completely genius play and
69.
by pertymcpert 6mo ago
The LLVM community used this model for years with Phabricator before it was EOL'd and moving to GH and PRs was forced. It's a proven model and works very well in complex code bases, multiple components and dependencies that can ha
70.
by pertymcpert 6mo ago
What might that be?
71.
by pertymcpert 6mo ago
I don't need any. I'm not making the claim that it's "most likely a lie".
72.
by pertymcpert 6mo ago
I'm not talking about Anthropic in particular. Other frontier labs will only be at most a year behind. I'm seeing the future here beyond just what's in front of us.
73.
by pertymcpert 6mo ago
Shit. Really? You mean they modified their frontier model to improve it and make it better and just called it a day? That their benchmarks which show step change improvements are just the result of successive changes on an EXISTING MODEL ?
74.
by pertymcpert 6mo ago
No one seems to have actually read the system card all the way through. The reason they didn't publish it was that it's orders of magnitude more successful at writing exploits vs Opus 4.6, which only managed it something like 2%
75.
by pertymcpert 6mo ago
Yeah...except Mythos's large context perf seems to be much better than Opus 4.6.
76.
by pertymcpert 6mo ago
Here: https://chatgpt.com/s/t_69d96c050078819199750da28ebd2526 I gagged after 2 sentences.
77.
by pertymcpert 6mo ago
Understood. I’m willing to defy guidelines and take the consequences. I still think it’s worth pointing out slop so people don’t waste their time talking to a machine.
78.
by pertymcpert 6mo ago
Each to their own. I appreciate you writing that comment yourself.
79.
by pertymcpert 6mo ago
I’m beyond trying to convince people to take this technology seriously. You’ll learn for yourself.
80.
by pertymcpert 6mo ago
If anything I’m seeing too much skepticism and not enough alarm. People burying their heads in the sand, fingers in their ears denying where this is all going. Unbelievable except it’s exactly what I expect from humans.
81.
by pertymcpert 6mo ago
What evidence makes you say that? Do you have insider info?
82.
by pertymcpert 6mo ago
This isn’t talking about compaction. This refers to performance as the model is loaded with 500k to 1m tokens.
83.
by pertymcpert 6mo ago
Did you read the article?
84.
by pertymcpert 6mo ago
They do care about files. They also care about how you express yourself, your tone, all sorts of seemingly unimportant details.
85.
by pertymcpert 6mo ago
I’d be very interested in trying it if you could spare the time to write up how to tune it well. If not thanks for the input anyway.
86.
by pertymcpert 6mo ago
That’s how the model is trained to signal the end to its generation and to indicate its thinking.
87.
by pertymcpert 8mo ago
There is a large body of research that shows it's not what you're saying it is FYI.
88.
by pertymcpert 8mo ago
Did you read the article?
89.
by pertymcpert 8mo ago
You need help.
90.
by pertymcpert 8mo ago
GCC and LLVM consider it a bug if the compiler is non-deterministic. If re-running the compiler generates different output because of things like address differences for example then it's something that needs to be fixed. So yes they

More ›