INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

cised

-0.76

ebted

-0.75

nesday

-0.73

rentices

-0.72

inous

-0.65

 weakest

-0.65

thy

-0.63

soDeliveryDate

-0.62

henko

-0.62

ulnerable

-0.60

POSITIVE LOGITS

netflix

0.79

 defenses

0.66

ographics

0.65

 complicity

0.63

ures

0.63

endants

0.61

 è£ıè

0.61

Catalog

0.60

RAM

0.60

 ILCS

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.