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

agne

-0.75

aez

-0.75

å¥³

-0.75

aceae

-0.74

meg

-0.72

adies

-0.70

chnology

-0.69

alys

-0.68

ãĤ¼ãĤ¦ãĤ¹

-0.67

umers

-0.67

POSITIVE LOGITS

he

1.32

his

1.13

 himself

0.93

his

0.92

He

0.88

He

0.86

His

0.86

him

0.82

His

0.80

he

0.75

Activations Density 0.000%

No Known Activations

This feature has no known activations.