INDEX

Explanations

references to individuals involved in violent incidents

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

0:0.05

1:0.04

2:0.05

3:0.05

4:0.05

5:0.13

6:0.03

7:0.03

8:0.05

9:0.11

10:0.14

11:0.21

Negative Logits

elvet

-1.65

obar

-1.64

uph

-1.41

agar

-1.39

ige

-1.38

pat

-1.37

 Houth

-1.36

 uniforms

-1.35

 Kirin

-1.33

cars

-1.32

POSITIVE LOGITS

 pond

1.63

ertodd

1.59

ulative

1.48

Brain

1.46

maxwell

1.41

average

1.38

 recovers

1.37

 stumble

1.35

 Played

1.35

ognitive

1.34

Activations Density 0.012%

references to individuals involved in violent incidents

No Known Activations