INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 environments

-0.72

correct

-0.70

ãĤ¼ãĤ¦ãĤ¹

-0.69

multi

-0.68

 environment

-0.68

urion

-0.65

amination

-0.65

anthrop

-0.64

tein

-0.63

assad

-0.63

POSITIVE LOGITS

 CrossRef

0.81

 Etsy

0.73

Zup

0.70

 PubMed

0.67

 Icar

0.67

NetMessage

0.67

 Olymp

0.66

 Unic

0.66

ãĢ

0.63

 Jugg

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.