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

JM

-0.83

BIP

-0.72

ockets

-0.71

inator

-0.70

HC

-0.63

lob

-0.62

 Croat

-0.62

ause

-0.61

 Macro

-0.61

ohydrate

-0.61

POSITIVE LOGITS

addons

0.86

atoon

0.71

andre

0.69

natureconservancy

0.68

 palace

0.68

basketball

0.68

 activ

0.67

akura

0.65

 Samurai

0.64

wrapper

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.