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

oÄŁ

-0.71

visors

-0.68

ypes

-0.67

rust

-0.66

éļ

-0.66

doms

-0.65

 Cosponsors

-0.65

ĪĴ

-0.64

 Bowser

-0.64

etooth

-0.64

POSITIVE LOGITS

aimon

0.66

 burner

0.64

 Caps

0.62

assi

0.60

 Testament

0.60

 album

0.60

IGN

0.60

 Finals

0.59

 sole

0.59

zac

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.