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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

rists

-0.67

vit

-0.65

 braces

-0.62

 icing

-0.62

 clues

-0.62

 vomit

-0.61

 scrolls

-0.61

boards

-0.60

 Tatt

-0.60

 lashes

-0.60

POSITIVE LOGITS

interstitial

0.95

closed

0.80

hower

0.78

ESCO

0.78

posed

0.77

olding

0.73

olded

0.72

OLD

0.72

Interstitial

0.72

 Niet

0.72

Activations Density 0.000%

No Known Activations

This feature has no known activations.