INDEX

Explanations

titles, headlines or text snippets containing capital letters and specific keywords or phrases

oai_token-act-pair · gpt-3.5-turbo

individual letters or symbols often used as initials or abbreviations

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

DragonMagazine

-0.66

coni

-0.63

poly

-0.63

 Generator

-0.62

 encyclopedia

-0.61

 Pengu

-0.61

 ILCS

-0.60

zbek

-0.59

 fung

-0.59

 fragmentation

-0.59

POSITIVE LOGITS

ained

1.34

ought

1.25

essed

1.24

oked

1.21

stood

1.21

ounced

1.19

aughed

1.15

owed

1.15

ocked

1.14

elled

1.12

Activations Density 0.131%

titles, headlines or text snippets containing capital letters and specific keywords or phrases

individual letters or symbols often used as initials or abbreviations

No Comments

No Known Activations