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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

UW

-0.73

 Luther

-0.66

 Croat

-0.64

 illum

-0.64

 Phar

-0.64

planet

-0.63

 indo

-0.61

Rw

-0.61

 libertarians

-0.60

 Holocaust

-0.60

POSITIVE LOGITS

ADRA

0.88

ierrez

0.76

wire

0.74

qus

0.72

Interstitial

0.72

uterte

0.71

CHAT

0.69

ENG

0.67

});

0.67

ais

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.