INDEX

Explanations

phrases related to yielding, submission, or bowing to authority

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L3_Hcat_z_lr1.20e-03_l19.00e-01_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v9.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.3.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

IFrame

Link

Head Attr Weights

0:0.02

1:0.03

2:0.08

3:0.15

4:0.02

5:0.03

6:0.05

7:0.10

8:0.08

9:0.22

10:0.07

11:0.10

Negative Logits

yssey

-1.18

ipop

-1.09

ivation

-1.08

��

-1.07

EQ

-1.06

utenberg

-1.03

uel

-1.02

ummer

-1.01

 setup

-0.99

uint

-0.98

POSITIVE LOGITS

龍契士

1.28

 Haram

1.21

gha

1.20

 modesty

1.14

��

1.13

 Pants

1.12

Accessory

1.12

galitarian

1.12

 traitor

1.12

 necks

1.10

Activations Density 0.003%

phrases related to yielding, submission, or bowing to authority

No Known Activations