INDEX

Explanations

names of countries and notable political figures

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

0:0.03

1:0.05

2:0.24

3:0.05

4:0.03

5:0.06

6:0.05

7:0.09

8:0.21

9:0.04

10:0.04

11:0.05

Negative Logits

chev

-1.48

Ps

-1.46

Yel

-1.41

<-

-1.37

amac

-1.28

uphem

-1.26

 "$:/

-1.25

nikov

-1.21

 Extras

-1.19

imentary

-1.18

POSITIVE LOGITS

 nexus

1.43

 fame

1.42

venge

1.33

 controversies

1.26

 reigning

1.24

 totality

1.20

 sponsoring

1.17

 sponsor

1.16

 regist

1.16

神

1.12

Activations Density 0.013%

names of countries and notable political figures

No Known Activations