INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

mma

-0.79

asma

-0.71

runners

-0.70

 Territories

-0.68

asons

-0.66

isco

-0.66

bourg

-0.66

 Malta

-0.65

utterstock

-0.65

ahime

-0.64

POSITIVE LOGITS

ãĤ¦ãĤ¹

0.75

 Flam

0.74

isine

0.68

 leveled

0.67

 Myers

0.65

cyclopedia

0.63

 liner

0.63

 piled

0.62

 signed

0.61

availability

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.