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

``

-0.88

GS

-0.70

ikes

-0.70

obe

-0.70

Taj

-0.68

hower

-0.67

ICE

-0.66

kees

-0.66

 ..."

-0.66

hess

-0.66

POSITIVE LOGITS

 blood

1.24

blood

0.92

Blood

0.91

 bodily

0.82

 Blood

0.81

 stabbing

0.77

 vigil

0.74

 griev

0.73

 tuition

0.67

 vascular

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.