INDEX

Explanations

code regex

np_max-act · gemini-2.0-flash

The neuron activates on regex‐specific metacharacters (backslashes, slashes, quantifier braces, anchors, etc.), effectively spotting bits of a regular expression.

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

	filename

-0.07

值

-0.07

NSNotification

-0.06

pac

-0.06

��이

-0.06

asket

-0.06

.expected

-0.06

erox

-0.06

hed

-0.06

umes

-0.06

POSITIVE LOGITS

^{}

0.07

ऊ

0.06

�

0.06

 проблемы

0.06

 programma

0.06

/"

0.06

 지방

0.06

ｭ

0.06

‚

0.06

Activations Density 0.013%

code regex

The neuron activates on regex‐specific metacharacters (backslashes, slashes, quantifier braces, anchors, etc.), effectively spotting bits of a regular expression.

No Comments

No Known Activations