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

DragonMagazine

-0.71

 STEP

-0.66

Streamer

-0.63

 Prelude

-0.61

TRUMP

-0.61

rette

-0.61

 Tweet

-0.61

roo

-0.60

EGIN

-0.59

itarian

-0.59

POSITIVE LOGITS

otin

0.85

gradation

0.72

odium

0.70

acca

0.65

arcer

0.64

ÃŃs

0.64

Ub

0.63

 belts

0.62

 cages

0.61

Armor

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.