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

 compr

-0.73

 supermarkets

-0.67

alde

-0.65

 dads

-0.63

isers

-0.61

theless

-0.60

ifference

-0.59

verty

-0.59

 overtake

-0.59

tsy

-0.59

POSITIVE LOGITS

=]

0.75

 shaft

0.69

Site

0.69

ILL

0.66

URI

0.65

Contents

0.63

DERR

0.62

ETHOD

0.62

utra

0.61

MC

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.