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

berra

-0.71

earchers

-0.70

ALSE

-0.69

rake

-0.68

wcsstore

-0.68

tern

-0.67

REM

-0.65

NPR

-0.65

ARC

-0.64

LAB

-0.64

POSITIVE LOGITS

 redistributed

0.66

 Venezuel

0.65

illery

0.63

 Ruin

0.62

ocrat

0.61

 blackmail

0.60

 bluff

0.60

 Provided

0.60

 Boxing

0.59

 Deposit

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.