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

ocene

-0.76

 Metatron

-0.69

FD

-0.63

WT

-0.62

 Morton

-0.61

kay

-0.61

pee

-0.60

 Frazier

-0.60

 Scroll

-0.59

osaurs

-0.58

POSITIVE LOGITS

robe

0.80

ighed

0.76

ebook

0.71

Drug

0.68

ettings

0.67

swers

0.67

oaded

0.66

Dri

0.64

itous

0.61

anguage

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.