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

Not in Any Lists

No Comments

Negative Logits

interstitial

-0.79

---------

-0.76

entric

-0.76

ends

-0.75

paralle

-0.75

esa

-0.72

auna

-0.70

lag

-0.68

instance

-0.67

âĢİ

-0.67

POSITIVE LOGITS

ye

0.70

 Cors

0.68

 Maxim

0.68

 Borough

0.66

 Chapel

0.63

 Enlightenment

0.63

 plum

0.63

 Ribbon

0.62

 PsyNet

0.62

 Marketable

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.