INDEX

Explanations

words related to identifying or marking something for a specific purpose or use

oai_token-act-pair · gpt-3.5-turbo

terms related to official classifications or designations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

corn

-0.72

Bey

-0.68

 Tide

-0.66

VIDEO

-0.66

ebook

-0.64

ears

-0.64

Roy

-0.64

Buzz

-0.64

conom

-0.63

Rh

-0.63

POSITIVE LOGITS

 designated

0.93

 warr

0.81

 designate

0.79

 satell

0.78

 Parenthood

0.76

 citiz

0.74

ignty

0.71

etheless

0.71

 designation

0.71

 naming

0.70

Activations Density 0.017%

words related to identifying or marking something for a specific purpose or use

terms related to official classifications or designations

No Comments

No Known Activations