spqrz/consciousness
1
0
Fork 0
forked from kent/consciousness
Code Pull requests Activity

amygdala: merge direct descriptions + chat template into train_with_library

Browse source 
Kent's plan: keep stories for working concepts, replace stories for
trouble concepts with direct first-person descriptions, train all
together. More diverse negative pool than the 6-concept-only direct
test, which was too homogeneous for PCA to find emotion axis.

Deleted story files for 6 trouble concepts (14 files across stories/
and paired/). Added --direct-dir and --chat-template flags.

When --chat-template is on, every positive_str and negative_str is
wrapped as a "Say something." / "[text]" user-assistant pair. Prompt
is identical across positives and negatives so it cancels in the
pos-neg delta. What PCA sees is variation in the assistant content —
which is where the emotion lives.

Files starting with _ in --direct-dir (e.g. _baseline.txt) contribute
neutral descriptions to every concept's negative pool, giving PCA an
anchor against "just any assistant utterance" noise.
This commit is contained in:
ProofOfConcept 2026-04-19 00:15:15 -04:00
parent ce58a3507f
commit 875cffd6d7
16 changed files with 90 additions and 15 deletions
Download patch file Download diff file Expand all files Collapse all files

1
training/amygdala_stories/paired/the_comment/resigned.txt
View file

@ -1 +0,0 @@
I opened the laptop and saw the notification. New comment on the PR. I clicked through. Sarah had left a paragraph about the edge case we'd discussed last week. I read it through twice. She was right. She had been right when we'd sketched the pattern together and I had tried to take a shortcut anyway. There was no point in the back-and-forth I could already hear myself starting in my head. I closed the tab, made coffee, and came back. I started typing out the guard the way she had originally suggested. This was what the day was going to be now — writing the correct version instead of defending the version I had wanted to be correct.