INDEX

Explanations

mentions of the name "Victorian."

oai_token-act-pair · gpt-3.5-turbo

references to the word "Victim" and related terms in a historical context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Nare

-0.63

 heck

-0.63

aq

-0.60

Jeb

-0.60

 Amos

-0.59

ËĪ

-0.59

 unfit

-0.58

ople

-0.58

 hybrids

-0.57

cov

-0.57

POSITIVE LOGITS

Vict

3.27

vict

1.16

 Victorian

1.15

 Victims

1.09

Sexual

1.03

 Victim

0.96

 Vict

0.92

 victim

0.92

Civil

0.90

Court

0.89

Activations Density 0.015%

mentions of the name "Victorian."

references to the word "Victim" and related terms in a historical context

No Comments

No Known Activations