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

-0.73

hani

-0.65

afety

-0.61

IPM

-0.61

 Agility

-0.61

Zup

-0.60

ibility

-0.59

 Stamina

-0.59

 Normandy

-0.58

 GOODMAN

-0.57

POSITIVE LOGITS

iant

0.68

arella

0.67

 lamps

0.67

»Ĵ

0.66

ificent

0.64

 caucuses

0.63

rogens

0.63

otomy

0.62

mble

0.61

 clot

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.