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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 HERO

-0.67

STON

-0.63

 assassins

-0.61

azon

-0.60

isher

-0.60

 Marriott

-0.57

 Ashe

-0.57

 outlet

-0.57

 clouds

-0.56

 Lynch

-0.55

POSITIVE LOGITS

 between

0.73

 until

0.72

between

0.72

til

0.71

atum

0.69

until

0.67

ngth

0.66

odynam

0.65

pperc

0.64

dayName

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.