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

dom

-0.81

zag

-0.74

raction

-0.74

chief

-0.72

urat

-0.71

cius

-0.69

unction

-0.68

competitive

-0.67

ango

-0.66

antis

-0.65

POSITIVE LOGITS

ships

0.73

jon

0.69

ANC

0.66

WARE

0.64

 Starfleet

0.64

 Educ

0.63

>]

0.63

iland

0.62

Disclaimer

0.62

 Spotify

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.