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

utsche

-0.76

aleb

-0.72

apego

-0.72

hovah

-0.65

legraph

-0.65

 psychiat

-0.65

legram

-0.64

usterity

-0.64

meet

-0.63

udic

-0.63

POSITIVE LOGITS

 flare

0.70

bourg

0.69

CLA

0.68

acs

0.62

sure

0.61

 plurality

0.61

CLE

0.60

OU

0.60

 refill

0.59

nova

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.