INDEX

Explanations

content sections and editing actions on a webpage

oai_token-act-pair · gpt-3.5-turbo

structured content and headings often found in a document or article

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

eg

-0.70

ee

-0.68

ted

-0.66

 asshole

-0.63

Pru

-0.63

Yas

-0.63

ulously

-0.63

 Greenberg

-0.61

 bliss

-0.60

Bey

-0.60

POSITIVE LOGITS

tenance

1.08

Contents

1.00

...]

0.81

erences

0.74

][

0.71

ä¹ĭ

0.70

chwitz

0.68

âĢ¦]

0.67

 charact

0.67

isode

0.66

Activations Density 0.023%

content sections and editing actions on a webpage

structured content and headings often found in a document or article

No Comments

No Known Activations