Feature #2258
Fires on “ an” (not labeled yet: read the examples)
- Fires on
- 0.53% of tokens
- Mostly in
- agent
- Layer
- after block 4
Where it fires
- agent 60%
- story 19%
- chat 9%
- python 7%
- worksheet 6%
Makes the model say
!"! little high. good and."
Makes it not say
ings<|arg|><|assistant_end|>denewString<|tool_start|><|assistant_start|>Rename
Strongest examples
On text the dictionary never trained on. The underlined token is where it fires most.
-
agent 2.92write, execute, memory<|system_end|><|system_start|>You are an AI agent running in
-
agent 2.9search, write, execute<|system_end|><|system_start|>You are an AI agent running in
-
agent 2.88search, write, execute<|system_end|><|system_start|>You are an AI agent running in
-
agent 2.88search, write, execute<|system_end|><|system_start|>You are an AI agent running in
-
agent 2.88search, write, execute<|system_end|><|system_start|>You are an AI agent running in
-
python 2.88]⏎```⏎⏎Bug: `has_item` should return True if x is in
-
agent 2.87search, write, execute<|system_end|><|system_start|>You are an AI agent running in
-
agent 2.87search, write, execute<|system_end|><|system_start|>You are an AI agent running in
-
agent 2.87search, write, execute<|system_end|><|system_start|>You are an AI agent running in
-
agent 2.86search, write, execute<|system_end|><|system_start|>You are an AI agent running in
-
agent 2.86search, write, execute<|system_end|><|system_start|>You are an AI agent running in
-
agent 2.86search, write, execute<|system_end|><|system_start|>You are an AI agent running in