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

elight

-0.67

 petrol

-0.65

 places

-0.64

)\

-0.64

 Dublin

-0.64

arious

-0.64

rium

-0.62

cliffe

-0.62

isphere

-0.61

 breached

-0.61

POSITIVE LOGITS

Assembly

0.69

veland

0.68

 Revolution

0.66

isconsin

0.66

 captcha

0.65

 captive

0.65

ournals

0.63

ãĥ£

0.63

Introduced

0.63

Revolution

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.