Toots for dachary@dacharycarey.social account

Written by Dachary on 2025-01-31 at 18:33

Run 3 results were worse than run 2. The LLM added more undesired categories, and categorized more things into these made-up categories.

I tweaked the prompt so I’m now feeding it one of three possible prompts, based on the code example programming language. I.e. JSON will never be a Usage Example, according to our definitions, so I have a version of that prompt that doesn’t include the category if the file is JSON.

Running again, then will add handling to reject invalid categories.

=> More informations about this toot | View the thread

Written by Dachary on 2025-01-31 at 12:35

Run 2 completed overnight! The results seem - worse - at a quick glance. The LLM hallucinated more categories this time, and assigned more examples to the hallucinated categories.

I think I’ll kick off another run without changing category definitions. Curious how the results will vary.

=> More informations about this toot | View the thread

Written by Dachary on 2025-01-31 at 03:43

We’re using an LLM to help assign categories to each code example based on some definitions we’ve come up with. We have over 20,000 code snippets across our docs corpus, so manually categorizing this volume just isn’t possible. We will spot check key product areas and any suspicious numbers (a Python snippet that came up as a CLI command, for example) and will take a reasonably accurate categorizing process as a starting point. Currently iterating on definitions for the LLM to improve results.

=> More informations about this toot | View the thread

Written by Dachary on 2025-01-31 at 03:38

For the curious, we are trying to categorize and count the code examples in our docs so we can get a better understanding of what we have. Developers ask for “more” and “more realistic” code examples, but we can’t really plan work to address gaps without more info. We have realized we have a very poor understanding of what already exists, so step one is to categorize and count our code examples by “type” and programming language. We can analyze an inventory of what exists to identify gaps.

=> More informations about this toot | View the thread

Written by Dachary on 2025-01-31 at 03:31

Ok! Based on my smaller sample sizes, I expected 5.5 hours to categorize these code examples, but it only took a little over 3.

Interesting results!

The LLM hallucinated 3 new categories during the process, but only assigned 1 snippet to each category, so I was able to easily spot-check those.

It performed poorly distinguishing the appropriate category for 1-line examples. I tweaked the category definitions, removed one category entirely, and 🤞 next run will improve. New run overnight!

=> More informations about this toot | View the thread

Written by Dachary on 2025-01-31 at 01:13

Glad I added that processed count! It’s nice to see it’s still going and over 4,000 examples processed.

=> More informations about this toot | View the thread

Written by Dachary on 2025-01-30 at 23:06

🤞

I just kicked off a program that uses a local LLM on my laptop to categorize code examples based on definitions we’ve come up with.

This ONE docs repo I’m working on (we have 50+) has 5,500 code examples. Based on some smaller data sets, it will take over 5.5 hours to complete this one repo, assuming it doesn’t crash.

I’ve got it outputting a processed count to the console every 100 files so I’ll wander by occasionally to see if it’s still running. Hope it doesn’t crash!

=> More informations about this toot | View the thread

Written by Dachary on 2025-01-20 at 13:14

Bed head.

=> View attached media

=> More informations about this toot | View the thread

Written by Dachary on 2025-01-14 at 13:54

Maui sits on my lap like this every single day. Lita has lived with us for ~11 years and this is the first time she has sat in my lap like this. She is trying out Maui’s spot!

=> View attached media

=> More informations about this toot | View the thread

Written by Dachary on 2025-01-07 at 14:06

Sigh.

No, colleague, PR review does not mean I will rewrite all your janky GPT-generated code so it actually works.

=> More informations about this toot | View the thread

Written by Dachary on 2025-01-02 at 14:30

WHY did I think it was a good idea to schedule a meeting for 9:30am the day I get back from a holiday, again?

🤦🏻‍♀️

=> More informations about this toot | View the thread

Written by Dachary on 2024-12-31 at 23:46

Tell me without telling me that you have won the genetic lottery and enjoy the privilege of being white and male.

=> View attached media

=> More informations about this toot | View the thread

Written by Dachary on 2024-12-30 at 18:58

I hit Perfection for the first time in Stardew Valley last night.

I started a new Playthrough today.

Apparently this game isn’t done with me yet…

=> More informations about this toot | View the thread

Written by Dachary on 2024-12-27 at 12:07

Up early today for a run out to Boston to drop off/pick up gifts for/from friends.

Set the alarm for 6am to feed the dogs, so car-sick dog hopefully digests his food enough to not barf it up as we cross the mountains.

Also, turn on the heat in the ambulance so it can get warm enough to not freeze us/the dogs on contact. Temp is -4 degrees F (-20 C) at time of waking so it’s gonna take a while to heat the ambulance.

Wish us luck!

=> More informations about this toot | View the thread

Written by Dachary on 2024-12-21 at 01:41

I keep seeing toots where I want to reply with something that starts with “Just…”

And then I just… don’t reply, instead.

It really is that easy.

=> More informations about this toot | View the thread

Written by Dachary on 2024-12-20 at 19:28

lol, oh noes! I have been playing with latte art - not seriously trying, just sort of doodling randomly with my steamed milk - and accidentally drew something obscene AND weirdly anatomically correct on my afternoon latte.

Gonna have to keep this sucker out of frame in the meeting I’m going into! (The danger of glass mugs.)

It did make me giggle tho 🤭

=> More informations about this toot | View the thread

Written by Dachary on 2024-12-18 at 02:00

BAD hair day? Or BEST hair day?

You decide.

=> View attached media

=> More informations about this toot | View the thread

Written by Dachary on 2024-12-17 at 14:26

Oh dear. I’ve reached the stage of not feeling well where somebody has set the status alert system to “Warn” and body is emitting a LOT of alerts related to feeling gross.

Might need to change my log rotation strategy because this is filling up my buffer FAST.

=> More informations about this toot | View the thread

Written by Dachary on 2024-12-11 at 15:54

Woohoo! My little Farming Simulator roleplay YouTube channel has finally hit 100 subscribers!

I’ve been making videos weekly for over a year - one a week at first, and then two a week for the last six months or so. This has been a very slow growth because it’s so incredibly niche. But I’ve been enjoying making them, so I’m just happy people are enjoying watching them enough to subscribe!

=> View attached media

=> More informations about this toot | View the thread

Written by Dachary on 2024-12-09 at 17:15

Every now and then, I order tea for wifey. I get her usual favorites, and try a few samples of new things she wouldn’t ordinarily buy for herself so she can try new things.

I am a coffee drinker. She is a tea drinker.

I thought we could try this “dark energy, herbal coffee” blend as a gateway of sorts between our preferred beverages.

Here’s her summary of our experience. (I agree, and would add that it was far inferior to the latte I made myself this morning.)

https://connectified.com/@masukomi/113623217903945101

=> More informations about this toot | View the thread

=> This profile with reblog | Go to dachary@dacharycarey.social account