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

 partName

-0.77

 Fargo

-0.64

 petrol

-0.64

 Relief

-0.64

 sclerosis

-0.63

 chim

-0.63

 cryptocurrency

-0.63

 reaction

-0.63

 clearance

-0.62

 tails

-0.62

POSITIVE LOGITS

asa

0.90

inelli

0.80

haw

0.80

iqueness

0.74

annon

0.74

ovies

0.72

aez

0.72

ove

0.72

ou

0.71

conservancy

0.71

Activations Density 0.000%

No Known Activations

This feature has no known activations.