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

 Canaver

-0.78

ãĥ¼ãĥ³

-0.78

awa

-0.78

pent

-0.73

Osw

-0.72

BuyableInstoreAndOnline

-0.70

dylib

-0.69

SAN

-0.69

TTL

-0.68

SOU

-0.68

POSITIVE LOGITS

 arranged

0.90

wcs

0.79

SPONSORED

0.70

dra

0.69

lis

0.68

annis

0.66

ract

0.66

Arab

0.66

 organized

0.63

arrow

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.