INDEX

Explanations

card

np_max-act · gemini-2.0-flash

This neuron activates on mentions of writing greeting cards (e.g., “birthday card,” “anniversary card,” “greeting card”)—i.e., it detects instructions requesting a card-style message.

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

.seed

-0.07

strcasecmp

-0.06

fire

-0.06

(check

-0.06

스트

-0.06

ises

-0.06

[mid

-0.06

 iliş

-0.06

.Attach

-0.05

Back

-0.05

POSITIVE LOGITS

.Sn

0.07

 Youth

0.07

(mu

0.06

<|begin_of_text|>

0.06

ạm

0.06

birthdate

0.06

ELLOW

0.06

 caption

0.06

 생각

0.06

Activations Density 0.011%

card

This neuron activates on mentions of writing greeting cards (e.g., “birthday card,” “anniversary card,” “greeting card”)—i.e., it detects instructions requesting a card-style message.

No Comments

No Known Activations