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

tracks

-0.87

estones

-0.83

players

-0.76

items

-0.74

estone

-0.74

ship

-0.71

Capture

-0.69

books

-0.68

iquette

-0.68

Attributes

-0.68

POSITIVE LOGITS

 Kinn

0.70

 revol

0.70

bably

0.69

 reluct

0.69

ITV

0.68

 unbeliev

0.68

 misc

0.67

 Ukrain

0.64

MIA

0.63

KKK

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.