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

VIDIA

-0.78

weights

-0.70

 Miko

-0.69

Haj

-0.68

agall

-0.68

ãĥĻ

-0.67

 Maiden

-0.66

è¦ļéĨĴ

-0.66

å§«

-0.66

 Devi

-0.65

POSITIVE LOGITS

 juven

0.70

ublic

0.66

alach

0.65

 incentiv

0.65

 cove

0.65

 scrut

0.63

 regard

0.63

 obliged

0.63

 satisfied

0.63

 inert

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.