INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

kinson

-0.87

bons

-0.78

-0.70

aeda

-0.69

ry

-0.68

 Salv

-0.65

illus

-0.65

channelAvailability

-0.64

driver

-0.64

zai

-0.63

POSITIVE LOGITS

 faded

0.69

 fade

0.69

 Doodle

0.68

 bask

0.67

 retreated

0.63

 occup

0.61

 digs

0.61

Anne

0.61

redo

0.60

dig

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.