INDEX

Explanations

phrases associated with controversial or polarizing political debates and identities

oai_token-act-pair · gpt-4o-mini Triggered by @bot

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

Not in Any Lists

No Comments

Head Attr Weights

0:0.02

1:0.01

2:0.06

3:0.33

4:0.01

5:0.02

6:0.07

7:0.11

8:0.06

9:0.11

10:0.05

11:0.09

Negative Logits

ngth

-1.45

ITNESS

-1.43

eele

-1.33

fty

-1.30

athered

-1.27

eller

-1.26

aternity

-1.23

ruary

-1.21

 Creat

-1.19

ellar

-1.17

POSITIVE LOGITS

 buffalo

1.18

 Bengal

1.17

 zombies

1.15

 psycho

1.15

 overdose

1.14

adesh

1.11

 altogether

1.10

 offence

1.09

 Sharks

1.09

 heroin

1.09

Activations Density 0.034%

phrases associated with controversial or polarizing political debates and identities

No Comments

No Known Activations