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

 helicop

-0.78

escape

-0.77

rane

-0.75

 explan

-0.75

 challeng

-0.74

 colle

-0.74

enum

-0.71

ĸļ

-0.71

 culminated

-0.69

 lighthouse

-0.68

POSITIVE LOGITS

Merit

0.77

Mah

0.63

 Voting

0.63

 Photographer

0.63

 Sultan

0.63

Mub

0.63

 Mour

0.61

actionDate

0.61

 Bund

0.61

Mig

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.