INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

áµ

-0.70

 Heavenly

-0.68

PokÃ©mon

-0.68

 Generations

-0.67

 Horizons

-0.66

 Falling

-0.65

Bin

-0.65

bies

-0.65

natureconservancy

-0.64

Fir

-0.64

POSITIVE LOGITS

 shut

0.73

pless

0.70

PG

0.70

tiny

0.70

yss

0.69

 Clarkson

0.68

TPS

0.68

apo

0.67

 amph

0.66

ampunk

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.