INDEX

Explanations

words related to warfare, especially nuclear warfare and related concepts like deterrence

oai_token-act-pair · gpt-3.5-turbo

references to nuclear and chemical warfare

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

este

-0.74

prints

-0.73

count

-0.71

ergy

-0.71

cerpt

-0.70

username

-0.69

otto

-0.69

val

-0.69

thumbnails

-0.67

zl

-0.67

POSITIVE LOGITS

 Warfare

1.21

 warfare

1.08

fare

0.96

riors

0.87

 hysteria

0.86

 posture

0.79

abwe

0.79

 mayhem

0.78

nesday

0.77

mallow

0.75

Activations Density 0.007%

words related to warfare, especially nuclear warfare and related concepts like deterrence

references to nuclear and chemical warfare

No Comments

No Known Activations