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

ydia

-0.77

ipes

-0.73

 externalToEVAOnly

-0.70

netflix

-0.70

quit

-0.66

ĨĴ

-0.63

=-=-=-=-=-=-=-=-

-0.62

acea

-0.62

oxin

-0.62

 vitro

-0.62

POSITIVE LOGITS

 awoken

0.73

ente

0.71

minster

0.70

bec

0.70

ansas

0.69

kay

0.66

Champ

0.65

 Rising

0.63

ISM

0.63

cand

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.