INDEX

Explanations

phrases or constructions indicating direction or attribution

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

1:0.01

2:0.04

3:0.06

4:0.15

5:0.02

6:0.03

7:0.35

8:0.03

9:0.03

10:0.09

11:0.12

Negative Logits

hunt

-1.59

ulia

-1.37

ebook

-1.37

[-

-1.33

ーティ

-1.32

 consumes

-1.30

▬

-1.28

ebin

-1.26

rowth

-1.25

cohol

-1.24

POSITIVE LOGITS

 Greenberg

1.51

 gesture

1.39

ohl

1.36

psey

1.33

 approving

1.31

aterasu

1.27

 lumin

1.25

hedon

1.24

 nods

1.21

 localization

1.20

Activations Density 0.001%

phrases or constructions indicating direction or attribution

No Comments

No Known Activations