INDEX

Explanations

negative descriptions or references related to the concept of "bad."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 5-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_5/width_16k/average_l0_68

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.5.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

WireFormatLite

-0.86

adaptiveStyles

-0.79

 mémor

-0.79

 Sunrise

-0.78

AndEndTag

-0.77

énario

-0.76

omock

-0.76

Sunrise

-0.74

 createSprite

-0.74

Dolphin

-0.73

POSITIVE LOGITS

bad

2.15

Bad

2.10

Bad

2.07

bad

2.02

BAD

2.00

BAD

1.79

 badly

1.06

 mauvaise

1.06

 mauvais

0.96

 schlechte

0.95

Activations Density 0.045%

negative descriptions or references related to the concept of "bad."

No Comments

No Known Activations