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

anqu

-0.80

acci

-0.76

asaki

-0.76

arters

-0.74

pur

-0.68

 resil

-0.67

Ń·

-0.67

schild

-0.64

 Majesty

-0.64

achi

-0.63

POSITIVE LOGITS

quickShipAvailable

0.74

utsche

0.61

 gamer

0.61

body

0.58

 Armenian

0.58

odd

0.58

 Ukrainian

0.57

nia

0.56

 grain

0.55

 Federation

0.55

Activations Density 0.000%

No Known Activations

This feature has no known activations.