INDEX

Explanations

pranks or pranking-related words or phrases

oai_token-act-pair · gpt-3.5-turbo

terms related to pranks and mischievous activities

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Twain

-0.81

farm

-0.79

ansas

-0.74

eva

-0.73

aeda

-0.73

Swe

-0.70

medi

-0.67

querque

-0.65

azy

-0.64

 Abortion

-0.61

POSITIVE LOGITS

 prank

0.95

iona

0.81

-+-+-+-+

0.79

 func

0.74

PIN

0.71

 mischief

0.70

Pry

0.70

obs

0.69

oun

0.66

orrow

0.66

Activations Density 0.008%

pranks or pranking-related words or phrases

terms related to pranks and mischievous activities

No Comments

No Known Activations