INDEX

Explanations

mentions of politicians, specifically focusing on the name patterns similar to "Biden" and "Clinton"

oai_token-act-pair · gpt-3.5-turbo

names related to political figures or identities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 exha

-0.89

vati

-0.77

rul

-0.75

pmwiki

-0.75

 weeds

-0.74

«ĺ

-0.73

 murd

-0.73

thora

-0.72

 ILCS

-0.72

ãħĭ

-0.71

POSITIVE LOGITS

iden

1.23

ovo

0.97

vier

0.92

ners

0.91

ception

0.89

unci

0.88

heimer

0.85

fold

0.84

vironment

0.83

ning

0.82

Activations Density 0.013%

mentions of politicians, specifically focusing on the name patterns similar to "Biden" and "Clinton"

names related to political figures or identities

No Comments

No Known Activations