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

)]

-0.79

=-=-

-0.73

@#&

-0.73

urry

-0.71

ombie

-0.70

âĻ¦

-0.70

uke

-0.67

olon

-0.67

 congr

-0.67

WARN

-0.66

POSITIVE LOGITS

 Coyotes

0.74

 arteries

0.71

 Penguins

0.71

sych

0.68

 Inqu

0.68

 enqu

0.66

 Islanders

0.65

 licens

0.64

izoph

0.63

 Inquiry

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.