INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

Eck

-0.70

ONSORED

-0.65

 Meier

-0.65

 Hail

-0.64

 Tall

-0.64

 Hann

-0.63

 Desire

-0.63

 Hazel

-0.62

 Polo

-0.62

 Giov

-0.61

POSITIVE LOGITS

obi

0.83

maxwell

0.80

prus

0.80

nces

0.78

merce

0.75

urring

0.75

akis

0.74

addons

0.74

ongyang

0.74

geries

0.72

Activations Density 0.000%

No Known Activations

This feature has no known activations.