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

 Coalition

-0.74

andum

-0.72

 PASS

-0.71

 Mata

-0.69

Alc

-0.65

 Advisory

-0.63

idth

-0.63

 Hogan

-0.63

 Hawth

-0.63

arten

-0.62

POSITIVE LOGITS

externalActionCode

0.72

bey

0.70

Wil

0.66

 discretion

0.66

 disposal

0.65

seller

0.64

 plent

0.64

 agric

0.63

 fingertips

0.63

vou

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.