INDEX

Explanations

government-related terms or titles

oai_token-act-pair · gpt-3.5-turbo

mentions of governmental and political figures and entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

terday

-0.80

ĻĤ

-0.74

toggle

-0.71

 captcha

-0.70

 traumatic

-0.67

anwhile

-0.65

 derail

-0.61

 moms

-0.61

 prol

-0.61

imeters

-0.61

POSITIVE LOGITS

 Accountability

1.24

 Himself

1.10

 Office

1.00

 Bureau

0.94

 Department

0.93

 Committee

0.92

 Agency

0.91

 Gazette

0.91

 Party

0.90

 Council

0.90

Activations Density 0.199%

government-related terms or titles

mentions of governmental and political figures and entities

No Comments

No Known Activations