INDEX

Explanations

phrases that indicate familial or relational connections

New Auto-Interp

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L5_Hcat_z_lr1.20e-03_l11.00e+00_ds49152_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v9.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

Data Type

float32

Hook Name

blocks.5.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

1:0.01

2:0.09

3:0.04

4:0.14

5:0.02

6:0.05

7:0.29

8:0.03

9:0.02

10:0.06

11:0.17

Negative Logits

opian

-1.43

oker

-1.36

 largeDownload

-1.35

anu

-1.33

ヴ

-1.30

inet

-1.26

negie

-1.26

lear

-1.25

rontal

-1.20

boards

-1.19

POSITIVE LOGITS

 Blumenthal

1.42

 retired

1.28

 former

1.23

 withd

1.23

 lazy

1.22

 unnamed

1.20

≡

1.18

 Cache

1.18

integer

1.13

 Handler

1.11

Activations Density 0.008%

phrases that indicate familial or relational connections

No Known Activations