INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 collar

-0.82

collar

-0.76

 actionGroup

-0.69

cyclopedia

-0.69

upper

-0.63

quickShipAvailable

-0.62

atell

-0.60

 jaws

-0.60

Hill

-0.60

ahoo

-0.60

POSITIVE LOGITS

Joined

0.68

 Sinn

0.68

ulhu

0.67

gdala

0.65

 Forsaken

0.64

 Alone

0.64

 Innocent

0.64

igans

0.62

 Gaul

0.62

ivation

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.