INDEX

Explanations

mentions of the word "Ad" followed by a number

oai_token-act-pair · gpt-3.5-turbo

references to advertisements or promotional content

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

OUP

-0.79

terday

-0.74

-0.72

ANGE

-0.71

externalActionCode

-0.69

ĸļ

-0.68

 Burk

-0.67

KEY

-0.65

Ń·

-0.64

 Kubrick

-0.62

POSITIVE LOGITS

vertis

1.33

vertisements

1.30

vertising

1.10

olesc

1.10

ept

1.06

mittedly

1.06

mit

0.98

olescent

0.98

hesive

0.96

elaide

0.96

Activations Density 0.006%

mentions of the word "Ad" followed by a number

references to advertisements or promotional content

No Comments

No Known Activations