INDEX

Explanations

LIMITED

np_max-act · gemini-2.0-flash

The neuron fires most strongly on tokens belonging to the standard file‐header/license boilerplate (e.g. comment markers, copyright years, “Google,” “Protocol Buffers,” macro guards, and license keywords).

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

_fill

-0.07

Ra

-0.07

�

-0.06

 Feed

-0.06

界

-0.06

关

-0.06

 이동합니다

-0.06

لفة

-0.06

อส

-0.06

OKEN

-0.06

POSITIVE LOGITS

propri

0.07

 ${↵

0.06

ских

0.06

power

0.06

Wid

0.06

recall

0.06

(SQL

0.06

anx

0.06

 щодо

0.06

Activations Density 0.005%

LIMITED

The neuron fires most strongly on tokens belonging to the standard file‐header/license boilerplate (e.g. comment markers, copyright years, “Google,” “Protocol Buffers,” macro guards, and license keywords).

No Comments

No Known Activations

LIMITED

The neuron fires most strongly on tokens belonging to the standard file‐header/license boilerplate (e.g. comment markers, copyright years, “Google,” “Protocol Buffers,” macro guards, and license keywords).

No Comments

No Known Activations