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

Phys

-0.77

ebus

-0.76

fu

-0.72

Gin

-0.70

æ©

-0.67

Ups

-0.64

TED

-0.64

 Whedon

-0.64

æ³

-0.63

 Tuls

-0.63

POSITIVE LOGITS

 Â®

0.95

Â®

0.91

âĦ¢

0.72

isphere

0.70

 licences

0.69

($)

0.68

 subsequ

0.67

isoft

0.65

itivity

0.65

iasco

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.