INDEX

Explanations

phrases related to attributes or characteristics enclosed in quotation marks

oai_token-act-pair · gpt-3.5-turbo

phrases involving quotations

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

Ļ½

-0.87

stant

-0.73

odi

-0.70

İĭ

-0.68

 Pengu

-0.66

¾

-0.65

icter

-0.64

 Evening

-0.62

ousse

-0.62

¸

-0.62

POSITIVE LOGITS

/"

1.14

 moniker

0.69

SPONSORED

0.65

aneers

0.64

>>\

0.63

Minecraft

0.62

OTUS

0.60

 remark

0.60

 excuse

0.60

AAP

0.59

Activations Density 0.082%

phrases related to attributes or characteristics enclosed in quotation marks

phrases involving quotations

No Comments

No Known Activations