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

 actionGroup

-0.75

 Flags

-0.68

 Stories

-0.68

Tube

-0.66

mobi

-0.65

 Cosponsors

-0.63

phia

-0.63

 Dimensions

-0.62

iframe

-0.62

 Trin

-0.62

POSITIVE LOGITS

minist

0.73

iency

0.70

aska

0.65

ient

0.65

cence

0.63

 obligation

0.62

nam

0.62

 Wildcats

0.61

etsu

0.61

innon

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.