INDEX

Explanations

proper nouns, particularly names of people

New Auto-Interp

Top Features by Cosine Similarity

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

0:0.03

1:0.01

2:0.04

3:0.05

4:0.04

5:0.04

6:0.43

7:0.09

8:0.05

9:0.06

10:0.06

11:0.05

Negative Logits

ERY

-1.50

 Janeiro

-1.40

BSD

-1.31

California

-1.22

▬

-1.19

UME

-1.19

 Proposition

-1.18

 Rite

-1.16

 Herb

-1.14

 Rica

-1.13

POSITIVE LOGITS

cknow

1.65

pal

1.60

rosso

1.57

rahim

1.45

ukes

1.42

slice

1.42

avy

1.30

lander

1.28

rays

1.25

iazep

1.24

Activations Density 0.001%

proper nouns, particularly names of people

No Known Activations