Feature #3438
Fires on “ write” (not labeled yet: read the examples)
- Fires on
- 0.51% of tokens
- Mostly in
- agent
- Layer
- after block 4
Where it fires
- agent 79%
- story 11%
- python 7%
- chat 3%
- worksheet 0%
Makes the model say
" python its welcome dog} from this
Makes it not say
slppatternouillioush
Strongest examples
On text the dictionary never trained on. The underlined token is where it fires most.
-
agent 3.75ent, webfetch, websearch, write, exec
-
agent 3.75ent, webfetch, websearch, write, exec
-
agent 3.73ent, webfetch, websearch, write, exec
-
agent 3.73ent, webfetch, websearch, write, exec
-
agent 3.72ent, webfetch, websearch, write, exec
-
agent 3.71ent, webfetch, websearch, write, exec
-
agent 3.71ent, webfetch, websearch, write, exec
-
agent 3.71ent, webfetch, websearch, write, exec
-
agent 3.7ent, webfetch, websearch, write, exec
-
agent 3.7ent, webfetch, websearch, write, exec
-
agent 3.69ent, webfetch, websearch, write, exec
-
agent 3.68ent, webfetch, websearch, write, exec