INDEX

Explanations

locations or places with specificity

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

0:0.09

1:0.03

2:0.07

3:0.15

4:0.02

5:0.11

6:0.02

7:0.02

8:0.23

9:0.09

10:0.08

11:0.03

Negative Logits

 iPads

-1.35

osaurs

-1.32

 Walsh

-1.29

 Carney

-1.28

 Sapp

-1.23

ravings

-1.22

 Vaugh

-1.20

itism

-1.19

 Lanka

-1.18

xes

-1.18

POSITIVE LOGITS

operated

1.40

majority

1.33

blooded

1.30

 practition

1.30

ennial

1.29

arist

1.27

ifter

1.25

ospons

1.22

centric

1.22

 derivative

1.22

Activations Density 0.540%

locations or places with specificity

No Known Activations