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

bid

-0.71

holding

-0.71

pak

-0.70

AAAA

-0.70

 Amendments

-0.69

sted

-0.69

sexual

-0.68

zb

-0.68

bol

-0.66

aer

-0.66

POSITIVE LOGITS

one

1.25

 ones

0.93

ription

0.77

 once

0.74

ONE

0.70

One

0.65

 oldest

0.65

 herald

0.64

 youngest

0.62

 Ones

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.