INDEX

Explanations

phrases related to deception and misleading information

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_16k/average_l0_71

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

明らかに

-0.48

EnableWeb

-0.46

 fourn

-0.45

bės

-0.44

 cimetière

-0.44

 couvert

-0.43

WebApi

-0.43

 remarquable

-0.43

DOCTYPE

-0.42

 couverts

-0.42

POSITIVE LOGITS

 mistakenly

1.65

 believing

1.60

 mistaken

1.54

 assumption

1.41

以為

1.41

 misconception

1.40

 erroneously

1.40

以为

1.39

 misunder

1.36

 wrongly

1.33

Activations Density 0.500%

phrases related to deception and misleading information

No Comments

No Known Activations