The @claudeai team usage data seems to think we had 46 pull requests in April.
In just ONE (of many) repos on the org connected to the account we have had 78 PRs.
I don't know where the breakdown is, but I haven't seen anyone else reporting how wrong these metrics are.
@vikasprogrammer@jameswlepage You can overload the context with unnecessary stuff if you register too many tools. More expensive and models perform worse.
The OpenAI Usage dashboard is almost unusable. Data and charts take 1 minute+ to load if they load at all.
So hard to get insights when you have bigger teams using lots of different models across a number of api keys.
@wjeffpetty In our town 9/10 league they can't lead off, but are allowed to steal 2nd or 3rd (no home) once the ball passes home plate. So most only go for it on passed balls. Seems like an ok balance to introduce the idea
Claude Mythos.
Ten trillion parameters: the first model in this weight class. Estimated training cost: ten billion dollars.
On the hardest coding test in the industry (SWE bench) it scores 94%.
It found a security flaw in a system that had been running for 27 years, one that every human engineer and every automated check had missed. It found another bug that had survived five million test runs over 16 years. (It did so overnight.)
It is so capable in cybersecurity that Anthropic will not release it to the public, instead it is launching Project Glasswing along with 100m in compute credits to help secure software.
Only twelve partners currently have access: Amazon, Cisco, Apple, Google, Microsoft, NVIDIA, JPMorgan Chase, Crowdstrike, Palo Alto, AWS, The Linux Foundation, Broadcom. (I'm sure the Pentagon is on the line?)
This is not a product launch: it is a controlled deployment of a system too powerful to distribute freely.
Tell me this isn't (very expensive) AGI?
@tvippolis Eh, there is an argument for some level skepticism of essentially every coaching hire that UNC has made in my lifetime other than Roy Williams. I can think of a number of them that seemed far riskier than this.
@BatchDude I like this compared to the realistic targets after Lloyd (ex: May was never real)
I'd 1000x rather roll the dice on Malone than Schertz or Byington. It's a big gamble, but I think the ceiling could be the highest of anyone.
Seeing this tactic more and more from people who consider themselves legitimate journalists. Intentionally getting facts wrong to inflate engagement because it increases the reach.
I wish the algorithms would bury this stuff because people cant help themselves from responding
What exactly happened here?
Overnight shipping order made on Tuesday afternoon
10am Wed email saying delayed
4pm email saying it shipped
7pm email saying delayed
Overnight means it is just coming from the local warehouse. So did the driver screw up?
1. His Dean Smith impression is funny, and it seems like they had a good relationship
2. Dean was 100% correct. The rules are the rules and you don’t change them to your liking. Hess getting “revenge” for being reminded of that just backs up everything I felt about him as a ref
@jaycaspiankang I got blocked by @tarheelblog for making that argument about Justin Jackson after the 2017 title. Some “fans” are just selfish and don’t actually care what is best for the players
@VicVijayakumar But you also need to run all the numbers. Estimated appreciation, primary residence capital gains exclusion from a sale (you have a 3 year window from when you moved), estimated rent/vacancy/maintenance, etc.
@VicVijayakumar I think it partially depends on where it is located.
We bought a new place 3 years ago. Kept the old one (2.5% rate w/8 years left) and are renting it out. Both in downtown Cary (hard to come by).
We may move back after kids are out of the house.
I hate systems that do this. It feels like the lazy approach to handling surges in loads and DB transactions.
Cary is not small, but it's not massive. ~180,000 people live here. The camps being registered for will have probably 10-100 kids each.