INDEX

Explanations

things that are physically connected or attached to something else

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate attachment or connection

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

audi

-0.79

geist

-0.74

lich

-0.72

RD

-0.70

iren

-0.70

vier

-0.68

iosity

-0.68

llor

-0.67

cens

-0.64

chlor

-0.64

POSITIVE LOGITS

 Siber

0.75

 orbit

0.73

 handle

0.73

 GPIO

0.72

 Equip

0.71

 shore

0.68

 Interstate

0.68

 assist

0.67

 underside

0.66

 schedule

0.66

Activations Density 0.174%

things that are physically connected or attached to something else

phrases that indicate attachment or connection

No Comments

No Known Activations