INDEX

Explanations

versions of words related to software interfaces

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

0:0.06

1:0.03

2:0.26

3:0.10

4:0.14

5:0.06

6:0.03

7:0.04

8:0.06

9:0.10

10:0.05

11:0.03

Negative Logits

rouch

-1.42

wcs

-1.41

bley

-1.41

gerald

-1.40

lington

-1.37

hap

-1.27

kefeller

-1.20

anmar

-1.20

nown

-1.19

negie

-1.19

POSITIVE LOGITS

ials

1.58

naires

1.48

ides

1.24

itude

1.13

umeric

1.13

lic

1.12

orce

1.12

IAL

1.11

ONE

1.09

 approval

1.09

Activations Density 0.002%

versions of words related to software interfaces

No Known Activations