INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 fault

-0.94

 Investig

-0.78

 Balk

-0.75

Wem

-0.74

 Fault

-0.72

 Americas

-0.71

 Euras

-0.71

 Univers

-0.70

 aber

-0.68

 Writ

-0.67

POSITIVE LOGITS

imal

0.97

alth

0.78

ks

0.75

robat

0.72

agy

0.72

ked

0.72

ebus

0.71

airo

0.70

icer

0.70

arning

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.