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

WINDOWS

-0.76

BuyableInstoreAndOnline

-0.73

nery

-0.73

IDA

-0.73

forum

-0.72

Display

-0.71

pload

-0.70

ITION

-0.69

TEXTURE

-0.68

åŃĲ

-0.68

POSITIVE LOGITS

 conductor

0.68

 Overt

0.67

imore

0.67

 striking

0.64

awed

0.63

weeney

0.62

 correspondence

0.62

 emerging

0.62

 ignition

0.61

erry

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.