INDEX

Explanations

references to various armed groups, especially those associated with terrorism

oai_token-act-pair · gpt-3.5-turbo

references to terrorist groups and their affiliations

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 etiquette

-0.71

nect

-0.69

 balcon

-0.69

 nutrit

-0.68

 clocks

-0.67

 unused

-0.67

Eck

-0.66

 electroly

-0.66

 wrinkles

-0.66

 restroom

-0.64

POSITIVE LOGITS

 Daesh

1.27

 Hezbollah

1.26

 ISIL

1.25

Nusra

1.24

 militants

1.23

 Taliban

1.22

 Islamic

1.22

 Islamist

1.19

 ISIS

1.18

Islamic

1.17

Activations Density 0.262%

references to various armed groups, especially those associated with terrorism

references to terrorist groups and their affiliations

No Comments

No Known Activations