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

²¾

-0.93

ĵĺ

-0.86

Ĥª

-0.81

agents

-0.79

etsk

-0.78

ļéĨĴ

-0.75

ķ

-0.73

assetsadobe

-0.72

antha

-0.70

tarian

-0.70

POSITIVE LOGITS

mount

0.68

 Graves

0.65

 Monroe

0.64

tails

0.60

illard

0.60

 Throne

0.60

isons

0.59

 Sons

0.59

 Dept

0.59

 Avalon

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.