INDEX

Explanations

references to psychological or physiological risks and related entities

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L9_Hcat_z_lr1.20e-03_l11.20e+00_ds24576_bs4096_dc1.00e-06_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.9.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

1:0.01

2:0.04

3:0.22

4:0.01

5:0.04

6:0.01

7:0.04

8:0.02

9:0.01

10:0.47

11:0.02

Negative Logits

 prints

-2.08

ilts

-2.07

 handwritten

-1.96

 typed

-1.93

 Uses

-1.88

 drawings

-1.87

 Donation

-1.86

shared

-1.85

 tweeted

-1.84

 Declaration

-1.83

POSITIVE LOGITS

angering

2.59

 squarely

2.54

 jeopardy

2.46

BuyableInstoreAndOnline

2.40

agame

2.36

 dangerously

2.18

 seriously

2.18

 disadvantage

2.16

agu

2.16

 reeling

2.14

Activations Density 0.060%

references to psychological or physiological risks and related entities

No Known Activations