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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ishers

-0.80

ringe

-0.70

ordinate

-0.68

enth

-0.68

ract

-0.67

isher

-0.66

zens

-0.64

idents

-0.64

romising

-0.63

ciplinary

-0.63

POSITIVE LOGITS

 http

0.79

ÃĥÃĤ

0.73

 https

0.72

www

0.69

 pione

0.69

 CLICK

0.67

ivia

0.63

 perspect

0.62

cloth

0.62

çĦ

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.