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

 pine

-0.76

 Sunny

-0.69

Pand

-0.65

Forest

-0.65

 Monsanto

-0.64

NAD

-0.64

 Waterloo

-0.63

 bamboo

-0.62

 hemp

-0.62

Sod

-0.62

POSITIVE LOGITS

stre

0.79

ibrary

0.75

/#

0.70

skill

0.69

aired

0.69

EStream

0.68

 Liberties

0.68

advant

0.68

itudes

0.68

repre

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.