INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

clud

-0.79

cott

-0.73

pron

-0.73

toe

-0.73

iris

-0.72

weed

-0.71

holding

-0.71

SEA

-0.71

alias

-0.70

aire

-0.69

POSITIVE LOGITS

 Heroic

0.76

 Shows

0.72

 Lunar

0.71

 Junction

0.70

ĨĴ

0.70

 Blazing

0.66

 Nost

0.64

 heart

0.63

onite

0.62

 Cance

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.