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

imal

-0.84

Ĳ

-0.82

ablish

-0.77

ormon

-0.76

ister

-0.76

Ĵ

-0.74

¿½

-0.72

omatic

-0.68

ı

-0.68

imate

-0.62

POSITIVE LOGITS

pas

0.67

 Parade

0.67

 Cadillac

0.65

 Biden

0.65

ties

0.64

 successors

0.63

Cic

0.61

leaders

0.61

swick

0.61

 cane

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.