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

ãĥ¬

-0.82

angles

-0.71

Pie

-0.71

 Ashes

-0.70

Contents

-0.68

KNOWN

-0.66

urn

-0.63

enty

-0.62

uned

-0.62

racted

-0.62

POSITIVE LOGITS

cius

0.75

owan

0.69

bda

0.66

 Candidate

0.65

 conjecture

0.64

osate

0.63

earcher

0.62

 Carroll

0.62

 guiName

0.62

 perpetrated

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.