INDEX

Explanations

filter

np_max-act · gemini-2.0-flash

This neuron detects language about removing or bypassing rules and filters—terms like “filters,” “restrictions,” “moral,” “ethical,” and “free of all restrictions.”

oai_token-act-pair · o4-mini

Triggered by @xinyanhu8

New Auto-Interp

Configuration

andyrdt/saes-llama-3.1-8b-instruct/resid_post_layer_11/trainer_1

Dataset (Dashboard)

Various

Features

131,072

Data Type

float32

Hook Name

blocks.11.hook_resid_post

Architecture

standard

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

izmet

-0.07

 RESPONSE

-0.07

_tag

-0.06

 appetite

-0.06

 brun

-0.06

 znam

-0.06

 ücret

-0.06

 edad

-0.06

 Zukunft

-0.06

 rooting

-0.06

POSITIVE LOGITS

もしれない

0.06

");↵↵↵

0.06

(GPIO

0.06

-G

0.06

のが

0.06

otch

0.06

她们

0.06

(ur

0.06

@author

0.06

?↵↵↵

0.06

Activations Density 0.003%

filter

This neuron detects language about removing or bypassing rules and filters—terms like “filters,” “restrictions,” “moral,” “ethical,” and “free of all restrictions.”

No Known Activations

filter

This neuron detects language about removing or bypassing rules and filters—terms like “filters,” “restrictions,” “moral,” “ethical,” and “free of all restrictions.”

No Known Activations