INDEX

Explanations

phrases discussing hazards and dangers related to various contexts and situations

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

Head Attr Weights

0:0.01

1:0.01

2:0.09

3:0.07

4:0.17

5:0.02

6:0.04

7:0.31

8:0.03

9:0.03

10:0.09

11:0.08

Negative Logits

ledged

-1.61

phabet

-1.56

atted

-1.53

ItemThumbnailImage

-1.52

riter

-1.51

itles

-1.49

endars

-1.47

agonist

-1.47

ermanent

-1.42

ragon

-1.38

POSITIVE LOGITS

 Communism

1.51

 Bridges

1.47

 Disclosure

1.47

 Zionism

1.43

 Cruel

1.40

 Dangerous

1.39

 Cornwall

1.39

 Kali

1.30

 exploitation

1.30

 piracy

1.29

Activations Density 0.010%

phrases discussing hazards and dangers related to various contexts and situations

No Known Activations