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

isSpecialOrderable

-0.71

 redes

-0.71

expr

-0.70

 discretion

-0.70

 antioxid

-0.69

tymology

-0.68

 totality

-0.65

 warranties

-0.65

 branding

-0.64

ĸļ

-0.64

POSITIVE LOGITS

 BALL

0.71

Sop

0.71

 Banner

0.69

achi

0.69

bestos

0.67

iencies

0.67

iences

0.66

 Probe

0.66

Hunt

0.65

itaire

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.