INDEX

Explanations

references to North Korea

oai_token-act-pair · gpt-3.5-turbo

mentions of North Korea

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Bron

-0.74

ocl

-0.73

âĸ¬âĸ¬

-0.72

Kop

-0.71

HER

-0.71

atin

-0.68

brand

-0.66

LOC

-0.66

UD

-0.66

 Walsh

-0.65

POSITIVE LOGITS

arthed

0.78

eln

0.77

orea

0.76

 Korea

0.75

 consulate

0.75

ongyang

0.74

 Pavilion

0.73

 embr

0.72

 Arabia

0.69

 cakes

0.69

Activations Density 0.033%

references to North Korea

mentions of North Korea

No Comments

No Known Activations