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

Person

-0.79

-0.77

person

-0.76

mal

-0.73

people

-0.73

Nob

-0.72

 "$:/

-0.71

woman

-0.70

hor

-0.69

 Michaels

-0.66

POSITIVE LOGITS

 shut

0.95

wagen

0.79

icago

0.77

oaded

0.70

enaries

0.70

ornia

0.69

ttle

0.69

ifax

0.66

nr

0.66

ysis

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.