INDEX

Explanations

references to system administrators or user roles in the context of system management

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of 'admin' and related terminology

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

enegger

-0.88

hovah

-0.85

False

-0.74

 Beir

-0.69

Temperature

-0.67

eele

-0.67

DonaldTrump

-0.66

tails

-0.66

 Empires

-0.66

 Winds

-0.66

POSITIVE LOGITS

stration

1.16

uthor

0.99

 admin

0.97

 administrator

0.89

 administ

0.88

isters

0.86

admin

0.85

ister

0.83

rative

0.82

 Admin

0.78

Activations Density 0.017%

references to system administrators or user roles in the context of system management

repeated mentions of 'admin' and related terminology

No Comments

No Known Activations