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

 Cosponsors

-0.75

 markup

-0.68

agle

-0.68

 turf

-0.67

 Fargo

-0.67

 Eggs

-0.66

doms

-0.66

manship

-0.65

 Corpus

-0.65

 Emmy

-0.61

POSITIVE LOGITS

onial

0.79

aepernick

0.77

 Clockwork

0.73

pport

0.68

iances

0.67

reon

0.65

llan

0.63

earable

0.63

akeru

0.62

================================================================

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.