INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

pora

-0.76

ulton

-0.67

asio

-0.67

worm

-0.67

 humidity

-0.66

iago

-0.66

 transporter

-0.64

hement

-0.64

antha

-0.64

udeau

-0.63

POSITIVE LOGITS

isSpecialOrderable

0.80

STER

0.78

AIN

0.78

soDeliveryDate

0.76

 Unsure

0.74

SPONSORED

0.72

[/

0.71

 Similar

0.70

VERTISEMENT

0.70

LIB

0.70

Activations Density 0.000%

No Known Activations

This feature has no known activations.