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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 crem

-0.71

ModLoader

-0.68

ington

-0.63

 licences

-0.63

 beet

-0.62

cess

-0.62

 Impossible

-0.61

etus

-0.61

 wastewater

-0.61

forth

-0.61

POSITIVE LOGITS

ATE

0.69

ATA

0.69

rug

0.68

ulator

0.66

ACA

0.66

 picks

0.65

 Reporter

0.65

å°Ĩ

0.63

rama

0.63

ippers

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.