Feature #3071
Fires on “ AI” (not labeled yet: read the examples)
- Fires on
- 0.18% of tokens
- Mostly in
- agent
- Layer
- after block 4
Where it fires
- agent 97%
- story 2%
- chat 0%
- python 0%
- worksheet 0%
Makes the model say
+ jump winCanved highair Can
Makes it not say
sgi4is new more asks
Strongest examples
On text the dictionary never trained on. The underlined token is where it fires most.
-
agent 3.22arch, write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode
-
agent 3.22arch, write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode
-
agent 3.22arch, write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode
-
agent 3.22arch, write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode
-
agent 3.21arch, write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode
-
agent 3.21arch, write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode
-
agent 3.19arch, write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode
-
agent 3.19arch, write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode
-
agent 3.18arch, write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode
-
agent 3.18arch, write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode
-
agent 3.18arch, write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode
-
agent 3.18arch, write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode