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

 Gems

-0.66

lings

-0.66

uras

-0.65

 EntityItem

-0.63

ones

-0.62

 Philippe

-0.61

ods

-0.59

itar

-0.58

 footing

-0.57

GO

-0.57

POSITIVE LOGITS

cffffcc

0.85

accompan

0.84

javascript

0.80

userc

0.77

ĵĺ

0.77

EStreamFrame

0.76

EStream

0.75

ļéĨĴ

0.73

amaru

0.72

 Shutterstock

0.72

Activations Density 0.000%

No Known Activations

This feature has no known activations.