INDEX

Explanations

references to knife-related incidents and violent encounters

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L10_Hcat_z_lr1.20e-03_l11.30e+00_ds24576_bs4096_dc1.00e-05_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.10.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

IFrame

Link

Head Attr Weights

0:0.37

1:0.05

2:0.05

3:0.03

4:0.05

5:0.02

6:0.05

7:0.11

8:0.02

9:0.04

10:0.07

11:0.08

Negative Logits

 Astros

-3.67

 Cosmos

-3.47

NXT

-3.44

 refuel

-3.42

 Bing

-3.34

 Morty

-3.28

 NASCAR

-3.23

 Atlantis

-3.20

////

-3.19

 Hilton

-3.19

POSITIVE LOGITS

 knives

8.15

 knife

7.70

 Knife

7.34

 blade

6.94

 blades

6.71

knife

6.26

 stabbing

6.24

 stabbed

5.92

Blade

5.90

 dagger

5.50

Activations Density 0.062%

references to knife-related incidents and violent encounters

No Known Activations