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

ctor

-0.75

 destro

-0.70

wcsstore

-0.68

 alle

-0.66

amate

-0.65

inventoryQuantity

-0.64

jri

-0.64

 trophies

-0.63

è¦ļéĨĴ

-0.63

CG

-0.62

POSITIVE LOGITS

knife

0.75

before

0.74

 before

0.71

 Breaker

0.68

 Hyde

0.68

toe

0.67

 Olsen

0.67

Jou

0.66

 Arnold

0.66

 Naked

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.