INDEX

Explanations

words related to espionage

oai_token-act-pair · gpt-3.5-turbo

references to espionage and spying

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

xual

-0.91

à¼

-0.75

ĸļ

-0.75

esville

-0.71

ŃĶ

-0.70

opathic

-0.70

clus

-0.68

orical

-0.67

urther

-0.65

usable

-0.65

POSITIVE LOGITS

sonian

0.95

glass

0.87

 spies

0.86

oleon

0.86

 spoof

0.83

spy

0.83

 spying

0.80

 satellites

0.78

OSH

0.75

ionage

0.71

Activations Density 0.018%

words related to espionage

references to espionage and spying

No Comments

No Known Activations