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

 socket

-0.71

leans

-0.69

irt

-0.68

pread

-0.67

Vert

-0.67

 discont

-0.67

ensibly

-0.65

mega

-0.65

 board

-0.63

yip

-0.63

POSITIVE LOGITS

 Kurdistan

0.79

amera

0.79

 Bulg

0.76

 Anthropology

0.75

-+

0.71

las

0.71

NRS

0.70

 Amph

0.69

finding

0.68

 Burnett

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.