Feature #3620
Fires on “ in” (not labeled yet: read the examples)
- Fires on
- 0.19% of tokens
- Mostly in
- agent
- Layer
- after block 4
Where it fires
- agent 97%
- story 1%
- chat 1%
- worksheet 0%
- python 0%
Makes the model say
this me liked liken play it f
Makes it not say
<|user_end|> !")caseratch"⏎ NA
Strongest examples
On text the dictionary never trained on. The underlined token is where it fires most.
-
agent 5.49write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode, a coding
-
agent 5.49write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode, a coding
-
agent 5.48write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode, a coding
-
agent 5.47write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode, a coding
-
agent 5.47write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode, a coding
-
agent 5.45write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode, a coding
-
agent 5.45write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode, a coding
-
agent 5.45write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode, a coding
-
agent 5.44write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode, a coding
-
agent 5.44write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode, a coding
-
agent 5.43write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode, a coding
-
agent 5.43write, execute<|system_end|><|system_start|>You are an AI agent running in OpenCode, a coding