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

IFrame

Link

Negative Logits

orate

-0.76

reon

-0.71

 Ezek

-0.66

redits

-0.66

Rat

-0.65

 invests

-0.63

ishment

-0.62

iliary

-0.62

 Moody

-0.62

ish

-0.62

POSITIVE LOGITS

VIDIA

0.81

CLASSIFIED

0.77

ettings

0.73

haar

0.70

ĺħ

0.68

 Loving

0.67

cano

0.64

 airflow

0.64

Ó

0.64

onymous

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.