INDEX

Explanations

imperatives and actions indicating movement or utility

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L5_Hcat_z_lr1.20e-03_l11.00e+00_ds49152_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v9.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

Data Type

float32

Hook Name

blocks.5.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.02

1:0.03

2:0.07

3:0.17

4:0.08

5:0.05

6:0.05

7:0.12

8:0.03

9:0.05

10:0.13

11:0.14

Negative Logits

rn

-1.43

)|

-1.40

eur

-1.40

ModLoader

-1.34

wp

-1.31

*)

-1.29

/)

-1.29

 Strongh

-1.28

女

-1.27

english

-1.26

POSITIVE LOGITS

 bypass

1.68

 potentially

1.48

 disconnect

1.47

 effectively

1.47

 diseng

1.46

 connect

1.45

 latch

1.45

 diagnose

1.44

 defer

1.43

 sever

1.42

Activations Density 0.037%

imperatives and actions indicating movement or utility

No Comments

No Known Activations