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

Balt

-0.70

ãĥ©ãĥ³

-0.70

wcsstore

-0.69

foundland

-0.68

>[

-0.68

Yen

-0.63

 Polaris

-0.62

soDeliveryDate

-0.62

%:

-0.60

RPM

-0.60

POSITIVE LOGITS

1.17

akespeare

0.81

 srfAttach

0.71

Âł

0.69

↵Âł

0.67

 prag

0.67

 Âł

0.66

______

0.64

 ÂłÂł

0.63

aryn

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.