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

ebus

-0.80

romy

-0.73

orius

-0.70

ibaba

-0.68

ijing

-0.67

 suite

-0.65

llah

-0.64

Zi

-0.64

elo

-0.64

 province

-0.63

POSITIVE LOGITS

natureconservancy

0.81

behind

0.76

*/(

0.74

IPM

0.73

duty

0.69

Dialogue

0.67

cd

0.64

history

0.64

0.63

CDs

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.