INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

 collaborators

-0.78

 compet

-0.70

 collaborations

-0.70

 imperson

-0.68

 veto

-0.68

 surrog

-0.64

orthy

-0.64

 coping

-0.63

 employers

-0.63

ompl

-0.63

POSITIVE LOGITS

edia

0.72

Console

0.72

chrome

0.71

aceae

0.70

 Vaugh

0.69

gradation

0.67

Companies

0.65

hol

0.65

Conservative

0.64

lehem

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.