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

BM

-0.66

olicy

-0.65

wid

-0.63

AMI

-0.63

 chast

-0.61

 Hels

-0.60

 Moder

-0.60

 Judgment

-0.59

":""},{"

-0.59

 terminals

-0.59

POSITIVE LOGITS

phabet

0.83

é¾įåĸļå£«

0.74

intestinal

0.71

omsky

0.71

 Dinosaur

0.70

checks

0.69

etch

0.69

tis

0.68

WAYS

0.68

 Dickinson

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.