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

leans

-0.76

etter

-0.75

erno

-0.72

ossom

-0.68

utra

-0.67

][

-0.65

icut

-0.65

lean

-0.64

ooming

-0.64

)|

-0.63

POSITIVE LOGITS

 Ministers

0.88

ŃĶ

0.78

 Administrative

0.73

Aust

0.73

Latest

0.70

arching

0.67

soDeliveryDate

0.67

Drag

0.66

 Planetary

0.65

soType

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.