INDEX

Explanations

mentions of people in positions of authority or leadership

oai_token-act-pair · gpt-3.5-turbo

references to positions of authority or command within organizations

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

umbnails

-0.75

aning

-0.69

 boredom

-0.68

 stink

-0.66

 definitions

-0.64

 FANTASY

-0.64

selves

-0.64

 selves

-0.63

 dances

-0.63

WAY

-0.62

POSITIVE LOGITS

 whom

0.79

 Misc

0.71

 Unch

0.71

chief

0.70

ilyn

0.68

ģĸ

0.68

 Kuwait

0.67

 Outer

0.65

 Haiti

0.65

 Sard

0.65

Activations Density 0.258%

mentions of people in positions of authority or leadership

references to positions of authority or command within organizations

No Comments

No Known Activations