INDEX

Explanations

names of locations or entities such as "Bend"

oai_token-act-pair · gpt-3.5-turbo

proper nouns, specifically names of locations and people

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 spoken

-0.74

 persuasive

-0.68

 readable

-0.66

 conclud

-0.66

ighting

-0.66

 dramas

-0.65

 drama

-0.63

 pricey

-0.63

 scaven

-0.62

 cruiser

-0.62

POSITIVE LOGITS

 Bend

1.20

SourceFile

1.01

stead

0.88

Rib

0.85

boro

0.84

oms

0.84

pole

0.82

olini

0.82

itude

0.81

owski

0.79

Activations Density 0.007%

names of locations or entities such as "Bend"

proper nouns, specifically names of locations and people

No Comments

No Known Activations