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

rices

-0.75

thumbnails

-0.74

 Subtle

-0.71

Appearances

-0.66

 Fake

-0.64

Gi

-0.64

Expl

-0.64

Rel

-0.64

pie

-0.63

imus

-0.63

POSITIVE LOGITS

 relegation

0.79

warts

0.68

ichick

0.67

 disbanded

0.67

 demoral

0.65

isson

0.64

 strugg

0.64

 apost

0.63

 disparate

0.63

 estranged

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.