INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

cks

-0.70

 Dwar

-0.66

 Dome

-0.65

emonic

-0.64

Tau

-0.64

Bei

-0.62

 Pound

-0.60

 Warfare

-0.60

king

-0.60

ifa

-0.59

POSITIVE LOGITS

soType

0.81

soDeliveryDate

0.77

mbuds

0.76

bats

0.76

afort

0.74

umenthal

0.73

inventoryQuantity

0.73

igmat

0.73

osponsors

0.71

catentry

0.70

Activations Density 0.000%

No Known Activations

This feature has no known activations.