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

ebook

-0.81

 exerc

-0.74

pez

-0.69

acts

-0.68

irtual

-0.68

ecast

-0.67

spection

-0.67

 Kard

-0.66

hatt

-0.65

heet

-0.65

POSITIVE LOGITS

sylvania

0.71

RESULTS

0.63

LLOW

0.62

ESE

0.62

FOR

0.61

 SHARES

0.61

LESS

0.60

Hz

0.59

ramid

0.58

 Kraken

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.