INDEX

Explanations

instances of exclusionary or derogatory language

New Auto-Interp

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

0:0.02

1:0.01

2:0.08

3:0.06

4:0.09

5:0.02

6:0.48

7:0.05

8:0.03

9:0.03

10:0.04

11:0.03

Negative Logits

 Beaut

-1.29

cknow

-1.23

MAS

-1.20

=-=-=-=-

-1.19

Phi

-1.16

 annexation

-1.16

 Tactics

-1.12

 istg

-1.12

 Summit

-1.10

 Championships

-1.10

POSITIVE LOGITS

erous

1.68

stood

1.57

than

1.52

 than

1.42

athed

1.38

hung

1.35

ptive

1.32

cientious

1.30

headed

1.30

 bothered

1.27

Activations Density 0.029%

instances of exclusionary or derogatory language

No Known Activations