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

alus

-0.75

 arsen

-0.67

 grips

-0.63

merga

-0.62

 melan

-0.62

ifice

-0.61

iencies

-0.61

 pains

-0.60

 scaff

-0.59

avascript

-0.58

POSITIVE LOGITS

 Generations

0.75

mobi

0.75

redo

0.73

 Ballard

0.68

)."

0.68

quel

0.68

android

0.67

active

0.63

vu

0.63

ettel

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.