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

 sqor

-0.80

habi

-0.76

 halftime

-0.71

merga

-0.67

 capit

-0.65

 antibody

-0.64

 membr

-0.64

ortmund

-0.63

 clos

-0.62

 antibodies

-0.62

POSITIVE LOGITS

Variable

0.74

 Dying

0.74

xp

0.69

Paper

0.66

Activ

0.62

Cy

0.62

 Vanity

0.62

0.61

tc

0.61

rek

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.