INDEX

Explanations

terms related to completion and processes

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L3_Hcat_z_lr1.20e-03_l19.00e-01_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v9.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.3.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

IFrame

Link

Head Attr Weights

0:0.02

1:0.03

2:0.09

3:0.04

4:0.02

5:0.04

6:0.04

7:0.07

8:0.17

9:0.20

10:0.06

11:0.17

Negative Logits

Magikarp

-1.15

SPONSORED

-1.15

iates

-1.12

 commenters

-1.12

 buggy

-1.12

arians

-1.10

 somet

-1.10

 bothered

-1.08

 alike

-1.08

AppData

-1.07

POSITIVE LOGITS

 groundwork

1.38

 bond

1.33

 probation

1.32

 duties

1.28

=~=~

1.26

 embargo

1.24

pection

1.23

ortment

1.23

 session

1.22

idelines

1.22

Activations Density 0.032%

terms related to completion and processes

No Known Activations