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

idis

-0.73

cules

-0.71

 chap

-0.68

 thous

-0.66

onica

-0.65

 Briggs

-0.64

stairs

-0.64

iverse

-0.62

 seiz

-0.62

oe

-0.61

POSITIVE LOGITS

antly

1.01

Ã¡s

0.70

llular

0.63

attled

0.62

 bottleneck

0.60

 processed

0.60

olf

0.60

GoldMagikarp

0.59

attering

0.59

WWE

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.