INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

Advanced

-0.78

ÃŁ

-0.71

orum

-0.69

geist

-0.69

 Feldman

-0.66

dict

-0.66

 Cellular

-0.66

xon

-0.64

uyomi

-0.64

Ga

-0.63

POSITIVE LOGITS

lia

0.76

orig

0.73

 cumbers

0.71

ient

0.69

fruit

0.68

vertisements

0.67

 helicop

0.66

 teasp

0.65

luck

0.65

 stuffing

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.