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

Negative Logits

ultimate

-0.72

riv

-0.68

scribed

-0.68

eed

-0.62

 Destroyer

-0.61

 choice

-0.60

vo

-0.60

-0.59

bered

-0.59

 Battles

-0.58

POSITIVE LOGITS

ËĪ

0.94

assian

0.84

 eyed

0.68

/+

0.65

elsius

0.65

Ã¡s

0.64

azel

0.64

aturday

0.63

ascus

0.63

taboola

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.