INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Elise

-0.71

Guest

-0.68

 Alexa

-0.64

 Mechdragon

-0.62

ãĥķ

-0.61

MPH

-0.61

 Brilliant

-0.61

 agenda

-0.60

ãĥķãĤ©

-0.60

>]

-0.59

POSITIVE LOGITS

ivo

0.90

alde

0.87

obal

0.78

isol

0.76

 Flavoring

0.76

ieft

0.76

oof

0.75

verty

0.74

thouse

0.74

ilater

0.73

Activations Density 0.000%

No Known Activations

This feature has no known activations.