INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 Kepler

-0.77

 Curve

-0.68

 Doodle

-0.66

 Expend

-0.65

 trillions

-0.64

 emot

-0.63

 custod

-0.63

doms

-0.62

 Burnett

-0.61

Vote

-0.60

POSITIVE LOGITS

meal

0.87

llah

0.83

uggage

0.76

alcohol

0.72

pa

0.71

apest

0.70

np

0.69

Äĩ

0.68

pill

0.68

 PACK

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.