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

nings

-0.75

Downloadha

-0.68

vin

-0.67

ãĥĵ

-0.65

 Kare

-0.63

 Gems

-0.62

nu

-0.60

intendo

-0.59

ãĥ³ãĤ¸

-0.59

ikuman

-0.59

POSITIVE LOGITS

fit

1.72

fit

1.16

pload

0.78

ileaks

0.77

Fit

0.75

Fit

0.70

 fits

0.69

hang

0.66

 Wilde

0.65

ready

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.