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

Wol

-0.76

Meh

-0.74

 Nare

-0.73

Beg

-0.71

acha

-0.67

åĬ

-0.63

Zam

-0.63

Nah

-0.62

Wid

-0.61

Yam

-0.60

POSITIVE LOGITS

NetMessage

0.80

ULTS

0.78

roups

0.74

agos

0.73

odynamics

0.71

mination

0.71

FAC

0.69

oven

0.67

ersen

0.67

ove

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.