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

etimes

-0.88

athering

-0.73

avior

-0.68

alf

-0.68

herer

-0.65

criminal

-0.65

thood

-0.65

ivals

-0.64

hea

-0.63

fighting

-0.63

POSITIVE LOGITS

éĹĺ

0.79

 Izan

0.75

~/.

0.72

ff

0.66

 Solitaire

0.66

 passer

0.65

bis

0.65

 Kindle

0.65

 Seasons

0.64

 Infinite

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.