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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 menace

-0.68

 Plague

-0.67

 robber

-0.67

 municipal

-0.62

 carnage

-0.62

idan

-0.62

 robberies

-0.61

 melancholy

-0.61

merce

-0.60

 Municipal

-0.60

POSITIVE LOGITS

Newsletter

0.86

Zup

0.86

OTO

0.80

Kay

0.73

 âĢº

0.72

Sche

0.71

Enlarge

0.70

 Rowling

0.69

OULD

0.68

hower

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.