INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 Properties

-0.65

 Inventory

-0.64

ajo

-0.63

kson

-0.62

Abstract

-0.60

Temperature

-0.59

 Ribbon

-0.59

enegger

-0.59

Entity

-0.58

 Handbook

-0.58

POSITIVE LOGITS

neg

0.78

 adolesc

0.78

Osw

0.71

ãĥ´ãĤ¡

0.70

âķĲ

0.68

udic

0.65

prone

0.63

trust

0.63

starting

0.61

 margins

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.