infryq

Yesterday I met with an ad hoc team working on a visualization web app for forecast evaluations, went to lunch early so I could make a 12pm meeting only to find it had been moved earlier, listened in on a webinar R&R gave on Delphi’s work and provided backchannel support in the chat, had a good 1:1 meeting with C, a reasonable team leads meeting (keeping people focused was a constant struggle but everyone’s clearly exhausted), and listened in on a couple final presentations for data analysis projects. In the cracks I started drafting documentation for the upcoming API signal on official hospitalization reports from Health and Human Services, and wrote up the release process for API signals in excruciating detail so that folks who are working while I’m away next week will be able to follow it. I’m glad I did; all wrapped up like that it makes it very clear how much work it is.

P did the Thursday errands, which included picking up the vegetable order, and this week picking up a painting I bought from R featuring vibrant colored crows ❤️🪶. It’s gorgeous and I’m excited to hang it. The roads were in good shape locally, though the farm folks had to do some substantial digging to get themselves out for deliveries. All told about 12”, wet and heavy.

Dinner, ordered a big box of Katie snacks, crossword, SG-1, snack, sleep. Dreamt of sewing a coat with many panels that kept skewing in ways that defied geometry, and making experimental puff pastry that alternated layers of butter and jam.

Finished Ring of Dragons. Endings continue to be of the got-tired-of-writing sort, but I sure am hoping to see the remaining threads come together harmoniously, so imma keep going.

Yesterday I had a cavity filled. I was unexpectedly nervous about it, but
everything went perfectly— no need for extra novocain, no surprises.

I worked from home the rest of the day just to keep an eye on recovery, but
everything went fine. I even found a github project for pretty much exactly
the thing we want to do. It’s a tensorflow implementation of a siamese*
model for text sequences— you train two lstms so that the encodings are
near each other for two strings referring to the same entity, and far when
the entities differ.

I did some spinning, and some knitting, including the last decrease round
for the waist shaping.

P had a very busy day, with lecture in the late morning, snow, coming home,
picking up the farm box, coming home, more snow, and lab in the evening.
The roads were pretty bad by the end of the day, but he got through it all
safely. We watched YouTube together over a late dinner, did the crossword,
and called it a night.

Remarkably nice day for a dentistry day.


*(ugh why)