Scott Ziolko

I have found this and asking to identify any unused or duplicate methods that can be removed to be helpful occasional prompts. If you aren't reviewing accepting/discarding changes fairly closely, I have found that the constant rewriting of files will insert new unnecessary functions.

This has changed recently. I rarely hit the limit, as keeping the chat very focused on a specific problem has been more successful, but recently, it hit the limit and offered to summarize the context to seed a new chat.

Minitab was the intro course since it had a gui and was students from lots of different majors so not programming oriented then splus on unix with the emacs speaks statistics addon or sas depending on the course and prof as this was pre-R

I've spent a minimal amount of time using continue.dev to use LLMs in the dev (and with local LLM models it all stays on my local env). It does well at describing what the SQL logic is doing (e.g. explain a model I didn't work on) but it doesn't seem great for other questions.

ah, the "photomoji" function on your phone will do the same isolation of the foreground object, I think for ios it is branded "memoji"

are they android users and you are on an iphone? it may be that they are using the google "photomoji" and instead of looking like a emoji reaction to you you are just getting a photo SMS

Also it is good pizza. The only challenge I have seen is that everyone orders then stares at their phone. Since they also use finger spelling to indicate when your order is ready people have to divert a little of their visual attention to looking for their order instead of listening for their number

At least for a while, I would expect that people will still find value in writing about new things. Something like how Twitter and Bluesky have been better at "now" than Google. People will publish and read about new stuff, but the latency of getting that into an LLM response will decrease.

But, they have legs which I was told was the great unsolved problem of VR

no, there haven't been any issues in orchestration. We did spend time to ensure that out of order and late arriving data is appropriately handled.

kubernetes

comment on things in a more public forum instead of the former more private walled garden that @dngrmike.bsky.social did not participate in

Also land doesn't vote... Oh a different misleading red map.

... and if you don't need the glasses they just might make things worse. Or if they are the wrong glasses you might see things that aren't there :)

Really great for churn dashboards and has also been a fruitful interview SQL discussion for data scientists. Working through how to do cumulative products for cohorts allows for a nice broad range of discussion topics!

or remove a couple layers since fine-tuning is not going to be as extensive as the pre-training

If you need to look up any of them they are here: eieio.games/blog/writing... @itseieio.bsky.social helpfully cataloged all of them

how do you all remember every UUID? I find it really hard. so I wrote them all down on every uuid dot com

the list has fast search across all 2^122 values (so you can find your favorites) - hoping to add some social features like "trending UUIDs" soon!

Let me tell you a story – one of the more haunting I’ve seen in 20 years of journalism. It’s about greed, death and denial. It took two years to unravel one doctor’s myth, a hospital’s complicity in creating it, and a documented trail of suspicious deaths…🧵

Mostly in agreement with @launchit.ai, but if I saw job postings for both at the same company I'd expect the architect to be more solving at the high level design of a complex system and the Staff/PE to skew towards solving the complex details in making the design work.

Assuming this is the initial review stage. 1) make sure that you have enough detail in the job description that applicants will know what skills are needed 2) have a well defined rubric for scoring resumes against #1. When done well you can also distribute the review work amongst your team.

I didn't see this mentioned, but "the Bible code" paper and follow-up analysis was an early case for me on the misapplication of stats methods side users.cecs.anu.edu.au/~bdm/codes/s...