INDEX

Explanations

references to wetness or being wet

oai_token-act-pair · gpt-3.5-turbo

references to wetness or moisture

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

Not in Any Lists

No Comments

Negative Logits

*/(

-0.73

å£«

-0.72

 distingu

-0.70

OSH

-0.67

 tremend

-0.67

ERO

-0.65

Requires

-0.65

########

-0.63

ONT

-0.62

DonaldTrump

-0.62

POSITIVE LOGITS

ting

1.37

ter

1.18

ted

1.15

ness

1.06

ters

1.05

lock

0.97

bag

0.97

nes

0.94

lands

0.92

nesday

0.89

Activations Density 0.023%

references to wetness or being wet

references to wetness or moisture

No Comments

No Known Activations