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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Constantin

-0.75

aten

-0.69

 Franco

-0.65

 Blanc

-0.64

ttp

-0.63

wd

-0.62

 Americ

-0.62

 Nept

-0.61

 dances

-0.60

 Demo

-0.60

POSITIVE LOGITS

MpServer

0.88

BIT

0.76

ãĤ®

0.68

ython

0.68

idge

0.67

erous

0.64

ãĤ©

0.63

rongh

0.63

paralle

0.62

é¾į

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.