INDEX

Explanations

image tags

np_max-act · gemini-2.0-flash

This neuron is primarily sensitive to the first numeric token that follows the “ond” fragment in those repetitive markdown image‐link sequences—i.e. it spots the leading numerical identifier in that “ond ### ##-000x” pattern.

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

 ticket

-0.07

 artists

-0.07

_coupon

-0.06

 adopt

-0.06

.neg

-0.06

 Pilot

-0.06

 suchen

-0.06

 ourselves

-0.06

 caric

-0.06

antis

-0.06

POSITIVE LOGITS

_COMMAND

0.07

лава

0.06

ithub

0.06

晴

0.06

�

0.06

agento

0.06

 جم

0.06

 ấn

0.06

vé

0.06

(payment

0.06

Activations Density 0.007%

image tags

This neuron is primarily sensitive to the first numeric token that follows the “ond” fragment in those repetitive markdown image‐link sequences—i.e. it spots the leading numerical identifier in that “ond ### ##-000x” pattern.

No Comments

No Known Activations

image tags

This neuron is primarily sensitive to the first numeric token that follows the “ond” fragment in those repetitive markdown image‐link sequences—i.e. it spots the leading numerical identifier in that “ond ### ##-000x” pattern.

No Comments

No Known Activations