INDEX

Explanations

tips and advice in various contexts

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

0:0.05

1:0.05

2:0.14

3:0.09

4:0.07

5:0.09

6:0.05

7:0.04

8:0.05

9:0.21

10:0.07

11:0.03

Negative Logits

 reconstructed

-1.31

uclear

-1.22

Unity

-1.18

 millions

-1.17

REAM

-1.17

-1.16

uilt

-1.15

 trillions

-1.13

 itself

-1.13

 situ

-1.13

POSITIVE LOGITS

 spam

1.45

 personalized

1.25

hooting

1.25

ockets

1.22

ommel

1.20

haw

1.19

 Readers

1.19

�醒

1.19

 reader

1.19

 lemon

1.18

Activations Density 0.013%

tips and advice in various contexts

No Known Activations