INDEX

Explanations

conduct/duct

np_max-act · gemini-2.0-flash

This neuron responds to any form of the word stem “conduct,” such as conduct, conducting, conductivity, conductive, etc.

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

Not in Any Lists

No Comments

Negative Logits

 swaps

-0.08

 Royal

-0.07

清

-0.07

 Pearl

-0.07

alf

-0.07

 sheep

-0.07

 zásob

-0.07

 evolutionary

-0.06

代

-0.06

 Simple

-0.06

POSITIVE LOGITS

 conduct

0.14

 conducting

0.13

 Conduct

0.12

 conducted

0.11

conduct

0.11

 conducts

0.11

 conductor

0.09

courses

0.08

 misconduct

0.08

、、

0.08

Activations Density 0.022%

conduct/duct

This neuron responds to any form of the word stem “conduct,” such as conduct, conducting, conductivity, conductive, etc.

No Comments

No Known Activations