INDEX

Explanations

proper nouns related to historical figures and geographic locations

oai_token-act-pair · gpt-3.5-turbo

references to the concept of a pontiff or pope

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

ITNESS

-0.87

-0.70

encers

-0.70

EStream

-0.70

encer

-0.69

berman

-0.67

ARE

-0.66

ggies

-0.66

oulder

-0.66

 VIDE

-0.64

POSITIVE LOGITS

Pont

1.21

 Pont

1.19

ificate

1.10

 pont

0.93

ified

0.92

unia

0.86

rovers

0.83

iac

0.83

ific

0.79

anus

0.79

Activations Density 0.005%

proper nouns related to historical figures and geographic locations

references to the concept of a pontiff or pope

No Comments

No Known Activations