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

 Maver

-0.73

 Martian

-0.65

 Bagg

-0.63

 Shattered

-0.62

anium

-0.61

Spawn

-0.61

idine

-0.61

 Canary

-0.61

 plun

-0.60

 smugg

-0.60

POSITIVE LOGITS

nl

0.74

lik

0.73

 partName

0.70

onse

0.69

pha

0.69

cms

0.68

olate

0.68

ongh

0.68

checked

0.68

\'

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.