Twitter | Pretraživanje | |
Janelle Shane 6. srp
Ravelry folks! In particular, crocheters! I have my first neural net-generated patterns posted to the LSG HAT3000 thread. Did it work? I have no idea. Some of the crochet notation in the dataset is VERY old, and it's not consistent at all. The results look like patterns maybe?
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 6. srp
Odgovor korisniku/ci @JanelleCShane
Here's another batch sampled from the same model, but with the top_k truncation set to 0 instead of 40. This allows the model to use lower-probability predictions. They are noticeably lower-probability.
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 7. srp
Odgovor korisniku/ci @JanelleCShane
It’s Hang in Wind! Humans cannot comprehend what it does.
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 7. srp
Odgovor korisniku/ci @JanelleCShane
Another batch of patterns, this time with truncation top_k = 60, a compromise that produces weird names but more-doable patterns than with truncation turned off.
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 7. srp
Odgovor korisniku/ci @JanelleCShane
If I don't give it a specific prompt, sometimes the neural net starts generating things that it remembers from its original training. I love its recovery method: twist the conversation back toward crochet hats.
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane
I’m not kidding about the all-conversations-lead-to-crochet-hats thing either. My prompt is the bit in bold. The neural net did the rest.
Reply Retweet Označi sa "sviđa mi se" More
Janelle Shane 7. srp
Odgovor korisniku/ci @JanelleCShane
Meanwhile Brim Hat Pattern #1708 is revealing itself to be an all-devouring eldritch horror
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 7. srp
Odgovor korisniku/ci @JanelleCShane
And if you can wear Crochet Cap Pattern #3615, you are already lost to the mortal realms
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 7. srp
Odgovor korisniku/ci @JanelleCShane
Here are some more hats from truncation top_k=60. I suspect that some of these don't result in hats? I probably should add more truncation to rein in the creativity a bit, but I'm enjoying these names so much.
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 7. srp
Odgovor korisniku/ci @JanelleCShane
Why do all the neural net's hats end up ballooning into universe-devouring monstrosities?
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 7. srp
Odgovor korisniku/ci @JanelleCShane
Decided to try a new method for inducing the neural net to generate interesting pattern titles: a conservative k=40 truncation but prompted with the phrase "Lord Vader!" Had the desired effect on the titles, as well as unintended effects on some of the patterns themselves.
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 7. srp
Odgovor korisniku/ci @JanelleCShane
oh no it's going to be space-devouring hyperbolic ripples again, isn't it? all of these patterns seem to be traps
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 7. srp
Odgovor korisniku/ci @JanelleCShane
tried to avoid hyperbolic chaos by reducing the sampling temperature from 1.0 to 0.8. now the neural net is playing it safer (arguably TOO safe in those 1st two patterns), but is it still going hyperbolic?
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 8. srp
Odgovor korisniku/ci @JanelleCShane
Update: these patterns totally are going hyperbolic. If anything, faster.
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 9. srp
Odgovor korisniku/ci @robin_h_p
Apparently exploding hats are nearly inevitable explains it well in the linked thread - each small excess increase ends up blowing up bc it's multiplied by later rounds. Meanwhile HAT3000 thinks it's still matching human patterns really well.
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 9. srp
Odgovor korisniku/ci @JanelleCShane
I trained HAT3000 for another 85 iterations to see if it might be able to fix its explodey hat problem. (here sampled with temperature 1.0, top_k 60, prompt text eccentric) The one called "The End..." looks dangerous even to me.
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 9. srp
Odgovor korisniku/ci @JanelleCShane
Tom Boy Pocket Bamboozles did not explode! It does have tentacles though
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 10. srp
Odgovor korisniku/ci @PaulaSimone
Meanwhile, Lumpy Top is another hyperbolic brain coral. calculates 1605 stitches in each of the outermost rounds.
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 11. srp
Odgovor korisniku/ci @JanelleCShane
Over 3500 stitches in one row. This hyperbolic neural net crochet pattern is now eating both yarn and sanity.
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 11. srp
Odgovor korisniku/ci @JanelleCShane
Results from a later training point (281 & 320). Prompting with eccentric text (Nobody expected stegosaurs in spaceships. etc) seemed to get induce it to generate interesting titles. Same temp=1 & top_k=60 I don’t know how it decided to try counting piglets instead of stitches.
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 21. srp
Odgovor korisniku/ci @JanelleCShane
Brim Hat Pattern #1708 is still devouring yarn. Starting to become mildly terrifying.
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 26. srp
Odgovor korisniku/ci @JanelleCShane
I exported a huge batch of unfiltered HAT3000 patterns - you can read the draft here. Note that some of the patterns appear to be, um, explicit. HAT3000 is a descendent of GPT-2, which Saw Some Shit on the internet. Prompt suggestions welcome
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 29. kol
Odgovor korisniku/ci @JanelleCShane
The larger GPT-2-774 model is released, but I don't have a way to finetune it on crochet yet. But I found that when I prompt it with a title + 2 lines of an existing crochet pattern, it has seen enough crochet on the internet that it already knows what to do.
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 4. ruj
Odgovor korisniku/ci @JanelleCShane
Lumpy top looks fabulous with red edge!
Reply Retweet Označi sa "sviđa mi se"
Janelle Shane 5. ruj
Odgovor korisniku/ci @JanelleCShane
I summarized the (faintly disquieting) story of HAT3000 here:
Reply Retweet Označi sa "sviđa mi se"