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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Flavoring

-0.83

clus

-0.76

ItemTracker

-0.74

COLOR

-0.73

cade

-0.72

atta

-0.72

adian

-0.71

schild

-0.70

 Kard

-0.70

ario

-0.69

POSITIVE LOGITS

ocate

0.71

 Surve

0.68

 videos

0.68

 Veter

0.65

 instincts

0.64

terday

0.63

 crawl

0.63

erers

0.61

 dist

0.61

 appra

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.