INDEX

Explanations

words related to legal issues, specifically fraud

oai_token-act-pair · gpt-3.5-turbo

references to fraud

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

%"

-0.69

ŃĶ

-0.67

âĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢ

-0.65

UTC

-0.62

Dak

-0.62

baugh

-0.62

 Arms

-0.60

bian

-0.59

 Candle

-0.59

 Views

-0.58

POSITIVE LOGITS

ulence

1.37

ulent

1.31

 fraud

1.02

sters

1.02

ster

0.99

 scam

0.86

ously

0.85

raud

0.84

 perpetrated

0.84

inals

0.78

Activations Density 0.010%

words related to legal issues, specifically fraud

references to fraud

No Comments

No Known Activations