INDEX

Explanations

names of specific individuals

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of specific names, particularly "Lyons" and "Stephens."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

itarian

-0.80

worthiness

-0.79

apixel

-0.73

ITY

-0.71

itized

-0.69

Kids

-0.69

efeated

-0.69

aido

-0.68

iov

-0.68

tor

-0.66

POSITIVE LOGITS

 Stephens

1.15

 Lyons

0.94

 slopes

0.81

terday

0.78

agher

0.75

anwhile

0.74

isson

0.72

 Barker

0.72

udence

0.72

 Greenwood

0.71

Activations Density 0.031%

names of specific individuals

repeated mentions of specific names, particularly "Lyons" and "Stephens."

No Comments

No Known Activations