INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

isations

-0.77

ALSE

-0.76

izations

-0.74

po

-0.68

 Carly

-0.64

TABLE

-0.64

 Chapters

-0.63

 accordingly

-0.63

isation

-0.63

 spells

-0.62

POSITIVE LOGITS

ĪĴ

0.77

onen

0.70

 bere

0.69

iru

0.67

imer

0.67

areth

0.65

 isEnabled

0.65

xtap

0.63

 bypass

0.63

ilts

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.