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

otin

-0.85

isle

-0.79

uku

-0.76

kefeller

-0.73

hess

-0.73

enhagen

-0.71

alus

-0.71

ograph

-0.70

eka

-0.70

gow

-0.68

POSITIVE LOGITS

Raz

0.82

 DRAG

0.74

 Razor

0.65

PDATE

0.65

ĻĤ

0.64

 excuse

0.64

 NIGHT

0.64

Ĥİ

0.63

©¶æ

0.63

 endeavors

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.