INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ĸļ

-0.70

 suffice

-0.65

icably

-0.63

 Orchestra

-0.62

 Inquiry

-0.62

 Muse

-0.60

XIV

-0.60

 Ideal

-0.59

Reviewer

-0.59

 Mania

-0.59

POSITIVE LOGITS

bing

0.72

ramid

0.70

alach

0.69

rosse

0.68

tes

0.65

phies

0.65

 brakes

0.65

Skin

0.65

ãĤ§

0.64

obin

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.