INDEX

Explanations

references to terrorist attacks

oai_token-act-pair · gpt-3.5-turbo

references to violent incidents or acts of aggression

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

OVA

-0.66

YC

-0.66

 Vale

-0.65

ETA

-0.65

 Wait

-0.65

view

-0.64

66666666

-0.64

 Vide

-0.63

VAL

-0.63

tz

-0.62

POSITIVE LOGITS

 attacks

1.24

attack

1.15

attacks

1.13

 attack

1.05

 Attacks

0.99

 attackers

0.99

Attack

0.89

 assaults

0.86

 spree

0.81

CVE

0.78

Activations Density 0.025%

references to terrorist attacks

references to violent incidents or acts of aggression

No Comments

No Known Activations