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

worm

-0.73

 Damascus

-0.70

iencies

-0.68

outer

-0.66

phis

-0.66

agne

-0.66

bys

-0.66

amiya

-0.64

itures

-0.63

orgetown

-0.63

POSITIVE LOGITS

 Consent

0.78

 reversible

0.74

ALSE

0.72

ARCH

0.69

¾

0.68

 consent

0.66

Ŀ

0.66

ARB

0.65

yll

0.65

ãĤ¼

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.