INDEX

Explanations

end

np_max-act · gemini-2.0-flash

This neuron activates on the substring “end,” i.e. the token “end” appearing (often as the start of words like end, endocrine, endoplasmic, endogenous, 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

 redundant

-0.06

 memberships

-0.06

ДА

-0.06

Bey

-0.06

");↵↵

-0.06

 whisky

-0.06

 народу

-0.06

 disappear

-0.05

 державної

-0.05

ilinear

-0.05

POSITIVE LOGITS

 kills

0.07

 core

0.07

 compiler

0.07

 Starts

0.07

 start

0.07

end

0.07

 Death

0.07

_Double

0.06

<Client

0.06

-end

0.06

Activations Density 0.020%

end

This neuron activates on the substring “end,” i.e. the token “end” appearing (often as the start of words like end, endocrine, endoplasmic, endogenous, etc.).

No Comments

No Known Activations