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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

dp

-0.74

oday

-0.73

iden

-0.73

oS

-0.72

eve

-0.72

 Curve

-0.67

daq

-0.66

 Akin

-0.65

 Tours

-0.62

Side

-0.61

POSITIVE LOGITS

gling

0.75

milo

0.67

mel

0.65

amping

0.63

heid

0.62

 watering

0.62

apple

0.62

gment

0.62

 Hungry

0.61

ulet

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.