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

ackets

-0.85

acus

-0.77

appa

-0.72

bent

-0.72

obbies

-0.71

gettable

-0.68

ornings

-0.68

rha

-0.67

neath

-0.67

agher

-0.66

POSITIVE LOGITS

Ingredients

0.82

ãĢİ

0.71

 SERV

0.68

Abstract

0.67

è¡

0.64

NESS

0.64

CLAIM

0.64

 Prosecutors

0.64

 Archdemon

0.63

Assembly

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.