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

 heel

-0.72

 unman

-0.69

addons

-0.68

hide

-0.66

rador

-0.66

 anten

-0.65

 station

-0.65

 Original

-0.63

 turf

-0.63

 embattled

-0.63

POSITIVE LOGITS

uci

0.80

STEM

0.78

\\\\\\\\

0.73

Ns

0.69

0.68

Americans

0.67

agos

0.67

atri

0.66

Charl

0.66

opter

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.