INDEX

Explanations

former positions or titles of individuals

oai_token-act-pair · gpt-3.5-turbo

references to individuals who have held previous positions of authority or prominence

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

achus

-0.73

opping

-0.70

uden

-0.68

morrow

-0.68

kun

-0.67

ofi

-0.67

omen

-0.66

uri

-0.66

ograph

-0.65

mand

-0.65

POSITIVE LOGITS

 Yugoslavia

1.12

 Yugoslav

0.94

 Soviet

0.93

 classmates

0.91

KGB

0.90

 classmate

0.89

 USSR

0.88

 employee

0.87

 colleague

0.83

 Presidents

0.83

Activations Density 0.054%

former positions or titles of individuals

references to individuals who have held previous positions of authority or prominence

No Comments

No Known Activations