INDEX

Explanations

words related to attitudes or outlooks

oai_token-act-pair · gpt-3.5-turbo

references to attitudes or opinions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

enegger

-0.96

idden

-0.83

icles

-0.82

icle

-0.80

 Printed

-0.79

locked

-0.75

bid

-0.72

anting

-0.72

hem

-0.71

Med

-0.70

POSITIVE LOGITS

 attitude

1.41

 attitudes

0.89

 towards

0.86

 toward

0.84

 conformity

0.80

yip

0.78

 stance

0.78

 indifference

0.76

ysis

0.75

 detector

0.75

Activations Density 0.008%

words related to attitudes or outlooks

references to attitudes or opinions

No Comments

No Known Activations