INDEX

Explanations

names or symbols with 'z', 'H', 'U', 'N', 'ai', or 'zz'

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L2_Hcat_z_lr1.20e-03_l11.00e+00_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v4.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.2.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

IFrame

Link

Head Attr Weights

0:0.07

1:0.03

2:0.13

3:0.03

4:0.28

5:0.06

6:0.02

7:0.03

8:0.08

9:0.13

10:0.05

11:0.03

Negative Logits

acists

-1.43

earchers

-1.41

osexual

-1.41

 undecided

-1.40

 millenn

-1.39

ModLoader

-1.38

Islam

-1.38

 moderates

-1.38

Pg

-1.37

itans

-1.37

POSITIVE LOGITS

acca

1.46

jer

1.33

 Malf

1.31

zz

1.31

 sacked

1.30

stra

1.29

heim

1.28

gar

1.28

ּ

1.27

cro

1.26

Activations Density 0.004%

names or symbols with 'z', 'H', 'U', 'N', 'ai', or 'zz'

No Known Activations