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

elsen

-0.75

 reckoning

-0.68

 disbanded

-0.65

asca

-0.63

 Manson

-0.63

-0.62

 crashed

-0.62

aldi

-0.62

 unbeliev

-0.61

DJ

-0.60

POSITIVE LOGITS

reprene

0.86

 Longh

0.78

Occ

0.76

ername

0.73

 Penny

0.70

yg

0.69

onom

0.67

uphem

0.66

Prem

0.66

ilateral

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.