INDEX

Explanations

mentions of positions of authority or leadership

oai_token-act-pair · gpt-3.5-turbo

references to positions of authority or leadership

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 IMAGES

-0.72

-0.70

 Mile

-0.64

 intercepted

-0.61

uder

-0.59

 dashed

-0.58

ILS

-0.58

adoes

-0.58

 Partnership

-0.57

 Adin

-0.57

POSITIVE LOGITS

 chair

1.36

chair

1.24

 Chair

1.19

 chairs

1.11

chairs

1.05

woman

1.04

lift

1.02

holder

1.02

Chair

0.95

person

0.87

Activations Density 0.008%

mentions of positions of authority or leadership

references to positions of authority or leadership

No Comments

No Known Activations