INDEX

Explanations

references to the United States or its abbreviation, USA, in the text

oai_token-act-pair · gpt-3.5-turbo

mentions of the United States

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

rog

-0.72

Wo

-0.69

 experimental

-0.65

 bridges

-0.65

 observational

-0.64

 hanging

-0.63

 complex

-0.63

 narr

-0.61

 marks

-0.61

 intersection

-0.61

POSITIVE LOGITS

USA

4.24

USA

2.13

usa

1.61

US

1.56

United

1.41

UK

1.39

Canada

1.30

America

1.26

Australia

1.26

SPA

1.17

Activations Density 0.012%

references to the United States or its abbreviation, USA, in the text

mentions of the United States

No Comments

No Known Activations