INDEX

Explanations

places, names of cities, and law enforcement-related terms

oai_token-act-pair · gpt-3.5-turbo

Triggered by @bot

references to police activity and investigations

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

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

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_post

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

NetMessage

-0.92

TPP

-0.89

å¥³

-0.85

 listened

-0.84

ãĥĻ

-0.83

ãĤ»

-0.81

ptions

-0.78

 survives

-0.77

ãĥ»

-0.76

washed

-0.75

POSITIVE LOGITS

 sheriff

1.10

 county

1.05

 coroner

1.01

 township

1.01

 municipality

0.98

 Greater

0.97

 city

0.97

 Humane

0.96

 Sheriff

0.95

 Metropolitan

0.95

Activations Density 0.225%

places, names of cities, and law enforcement-related terms

references to police activity and investigations

No Known Activations