INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 Bucket

-0.71

MK

-0.69

 Gundam

-0.67

ably

-0.67

 Gott

-0.66

 Totem

-0.65

 Vert

-0.65

witz

-0.64

APR

-0.62

 encoded

-0.61

POSITIVE LOGITS

userc

0.84

andise

0.83

past

0.74

unal

0.70

Govern

0.70

cffff

0.69

OTAL

0.68

FORMATION

0.67

 scient

0.65

afort

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.