INDEX

Explanations

terms related to espionage and spying

oai_token-act-pair · gpt-3.5-turbo

references to spies or espionage

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

à¼

-0.87

xual

-0.83

ĸļ

-0.78

esville

-0.77

haust

-0.71

usable

-0.69

verend

-0.69

orical

-0.67

oric

-0.67

veyard

-0.67

POSITIVE LOGITS

 spies

0.88

sonian

0.87

spy

0.84

oleon

0.82

 spoof

0.80

OSH

0.78

 satellites

0.76

glass

0.74

wat

0.71

 transmitter

0.70

Activations Density 0.009%

terms related to espionage and spying

references to spies or espionage

No Comments

No Known Activations