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

Negative Logits

 Optimus

-0.68

 Unic

-0.66

 Album

-0.64

 berth

-0.63

 Packers

-0.62

 Meadow

-0.62

ãĥŁ

-0.62

 Ideal

-0.61

MX

-0.61

 Ireland

-0.61

POSITIVE LOGITS

 tremend

0.83

sed

0.81

ó

0.78

omore

0.75

 spokes

0.73

imir

0.72

heastern

0.67

achus

0.67

ando

0.66

ingu

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.