INDEX

Explanations

phrases indicating lists or selections

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 4-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_4/width_16k/average_l0_124

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.4.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

QMetaType

-0.76

StructEnd

-0.76

 Monfieur

-0.73

ustimmung

-0.72

MigrationBuilder

-0.70

 Houſe

-0.68

 itſelf

-0.66

 Jefus

-0.64

----</

-0.64

 препратки

-0.63

POSITIVE LOGITS

respective

0.60

hichever

0.59

ITHER

0.57

いずれ

0.56

 either

0.55

zości

0.54

staande

0.53

zość

0.53

 Eisenberg

0.53

 métiers

0.53

Activations Density 0.159%

phrases indicating lists or selections

No Comments

No Known Activations