INDEX

Explanations

words related to official positions in organizations or institutions, such as "Department" or "Police Department spokesman"

oai_token-act-pair · gpt-3.5-turbo

mentions of government and law enforcement entities

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

iphate

-0.79

isable

-0.77

interstitial

-0.75

itaire

-0.70

burst

-0.67

ilst

-0.66

...]

-0.65

âĢ¦]

-0.64

phe

-0.63

imation

-0.62

POSITIVE LOGITS

 officials

1.55

 spokesman

1.50

 spokeswoman

1.46

 spokesperson

1.36

 spokes

1.33

 staffers

1.28

 Deputy

1.22

 aides

1.19

 insiders

1.19

 Chairman

1.16

Activations Density 0.193%

words related to official positions in organizations or institutions, such as "Department" or "Police Department spokesman"

mentions of government and law enforcement entities

No Comments

No Known Activations