INDEX

Explanations

No Explanations Found

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

PHI

-0.67

volt

-0.64

REG

-0.63

cision

-0.61

clips

-0.60

tin

-0.60

aster

-0.60

Math

-0.60

chell

-0.59

tips

-0.59

POSITIVE LOGITS

aeda

0.86

 unden

0.70

ysical

0.68

eworld

0.65

uit

0.65

isen

0.63

 Kerry

0.63

alos

0.63

alty

0.63

 confir

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.