INDEX

Explanations

references to notable individuals and their roles or affiliations

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

0:0.08

1:0.02

2:0.11

3:0.04

4:0.05

5:0.05

6:0.16

7:0.06

8:0.17

9:0.15

10:0.03

11:0.04

Negative Logits

 Aram

-3.89

gaard

-3.57

 Zoro

-3.24

Sod

-3.20

ordan

-3.19

 Jordanian

-3.08

 latex

-3.07

 Kuwait

-3.06

 hooked

-3.02

 condem

-3.01

POSITIVE LOGITS

 McCabe

10.54

McC

6.70

 McAuliffe

5.83

 McCoy

5.08

 McCain

4.76

McA

4.51

McC

4.34

 Torch

4.33

 Comey

4.32

 Flake

4.16

Activations Density 0.000%

references to notable individuals and their roles or affiliations

No Known Activations