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

Not in Any Lists

No Comments

Negative Logits

ngth

-0.86

 positional

-0.74

itant

-0.73

natureconservancy

-0.72

UES

-0.70

ãĤ¼ãĤ¦ãĤ¹

-0.67

enium

-0.67

llah

-0.65

emonic

-0.63

 compos

-0.61

POSITIVE LOGITS

etheless

0.74

Tro

0.71

Commerce

0.68

 Advertisement

0.66

ãĥ©ãĥ³

0.66

Sec

0.64

arta

0.63

>>\

0.62

merce

0.62

jar

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.