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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

atre

-0.81

ravel

-0.68

rill

-0.68

 tremend

-0.67

bent

-0.66

ilib

-0.65

irie

-0.64

bris

-0.64

yrinth

-0.64

Rex

-0.64

POSITIVE LOGITS

hips

0.90

 Nadu

0.79

hip

0.73

 ACPI

0.63

 Elias

0.62

ylon

0.61

Äį

0.60

 sweat

0.58

 Relative

0.58

teenth

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.