INDEX

Explanations

mentions of fake news

oai_token-act-pair · gpt-3.5-turbo

references to fake news

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

xual

-0.85

Marginal

-0.73

pai

-0.73

hem

-0.72

endez

-0.69

night

-0.69

Reviewer

-0.68

served

-0.67

waukee

-0.66

 sqor

-0.66

POSITIVE LOGITS

 news

1.00

IDs

0.96

ument

0.96

 NEWS

0.87

 identities

0.81

pas

0.80

 positives

0.80

tails

0.80

 News

0.78

news

0.73

Activations Density 0.074%

mentions of fake news

references to fake news

No Comments

No Known Activations