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

ĸļ

-0.91

ļéĨĴ

-0.85

Ń·

-0.81

natureconservancy

-0.76

iqueness

-0.72

 Cruel

-0.71

 Daredevil

-0.69

Cla

-0.68

 Thrones

-0.68

Dra

-0.67

POSITIVE LOGITS

ingham

0.80

insured

0.76

HP

0.70

agic

0.69

imb

0.67

ask

0.67

ope

0.67

atic

0.66

ussen

0.66

onomous

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.