INDEX

Explanations

phrases that emphasize personal connection and vulnerability, particularly through eye contact

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

0:0.01

1:0.04

2:0.06

3:0.03

4:0.01

5:0.06

6:0.03

7:0.06

8:0.05

9:0.46

10:0.07

11:0.06

Negative Logits

ovember

-1.59

rosse

-1.24

Camp

-1.22

angan

-1.22

 campaigned

-1.20

inqu

-1.18

unch

-1.17

ipl

-1.16

Stretch

-1.14

igun

-1.14

POSITIVE LOGITS

 positives

1.51

 VIDEOS

1.43

 directions

1.39

 negatives

1.38

assian

1.36

QR

1.34

 Indies

1.33

 Companies

1.32

 blinking

1.30

flix

1.27

Activations Density 0.163%

phrases that emphasize personal connection and vulnerability, particularly through eye contact

No Comments

No Known Activations