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

CVE

-0.86

chwitz

-0.80

 Mubarak

-0.74

 coughing

-0.71

nia

-0.67

 Rollins

-0.67

journal

-0.67

 Copenhagen

-0.66

 respir

-0.66

lihood

-0.66

POSITIVE LOGITS

 Adventure

0.71

atl

0.66

bor

0.66

Asc

0.65

eways

0.65

ugu

0.63

aiden

0.63

atoes

0.62

alities

0.61

Kin

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.