INDEX

Explanations

numerical values in a specific format, possibly related to pages or sections of a document

oai_token-act-pair · gpt-3.5-turbo

specific numerical data and monetary values

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

•Many-Token Promoting Features

Space + All Caps

No Comments

Negative Logits

aily

-0.82

cephal

-0.78

ciplinary

-0.78

ascript

-0.76

achev

-0.73

ogan

-0.73

gebra

-0.71

sie

-0.70

unal

-0.70

aber

-0.70

POSITIVE LOGITS

WA

1.35

HI

1.32

MI

1.32

FL

1.31

DE

1.29

SC

1.28

SEA

1.27

FR

1.26

ON

1.25

NET

1.25

Activations Density 0.284%

numerical values in a specific format, possibly related to pages or sections of a document

specific numerical data and monetary values

No Comments

No Known Activations