omg! 🤍

Attack of a tiny white dog

AHHHHHHHHHHHHHHHHHHHHHH

seating installed

merry Aldi cheese advent calendar to you

so cute I wanna bunt it

curly slurper they call me

go look at Shayne Taylor's murals on the kayak storage boxes at River Park oh my god

double dog duo*

other double dog set doing pretty well too

it works really well and they kind of pull each other along

but the likelihood Bernie pees on Cinder has significantly increased lol

I don't know the full details of the specific mowed spot but yeah monocultures of smooth brome and tall goldenrod have been an issue at the site for a long time. There's some good stuff too but relatively low diversity & the ironweed grows like 10' tall.. not compatible w/ growing new baby natives 🌱

finally testing a leash splitter for the first time

it was seeded this spring

hi vis vest lol

my mom 🏍️ biked 🏍️ here from Duluth

I don't know what gives you that impression but it doesn't seem like you're assuming good faith and these are amazing people on the iNat team so I'm done here

I don't think they've said that new text would be generated every time? They really haven't said much at all because it's so early in the project.

A few tips could for example be generated for each taxon in the computer vision model (only 100k spp) and then only updated occasionally or on demand.

What is it that you are expecting more than this? Keeping in mind it's a demo project that doesn't even exist yet?

that's why I said "for what it's worth"

their grant doesn't require them to even work wth Google infrastructure

From the blog post update: "we have no obligation to use Google’s infrastructure as part of this grant"

fwiw their current CV training machines are run on renewable energy

The point is - we don't know what or who they will work with.

e.g. last year Microsoft provided "generous support" and they (with university partners) built a demo vision language model based off hugging face architecture.

ya when it's ticky I put my clothes tick quarantine rubbermaid w/ permethrin (no 🐈)

that drive home though ☠️

They have not said anything about using existing external models. And their blog post calls out that all their existing work on computer vision is done in-house.

You can use the accuracy circle: manually move the map pinpoint a bit and still include the true location within the circle. Some people do that instead of obscuration, especially home addresses.

Also as someone with tens of thousands of edits on Wikipedia pages about plants and animals, it's interesting how many people are commenting asking for a wiki.

That platform exists, come help edit.

I'll help if you have questions about how species identification info can be best incorporated.

I'll probably end up deleting this because I don't really want to get yelled at on the internet, but negative comments are always far out of proportion of actual user base sentiment, so felt the need to say something.

I know not everyone realizes it's a labor of love & the deep respect for nature/ conservation that iNat staff have, many of which have been there since the beginning (including the creator). But it's disappointing to see the misconceptions, fearmongering, & shortsightedness by deleting accounts.

Re: iNaturalist AI backlash. I have high trust and confidence in the iNaturalist staff to review community feedback and move forward with a path that makes sense.