INDEX

Explanations

words related to legal and bureaucratic procedures

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L2_Hcat_z_lr1.20e-03_l11.00e+00_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v4.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.2.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

IFrame

Link

Head Attr Weights

0:0.04

1:0.03

2:0.28

3:0.05

4:0.24

5:0.04

6:0.03

7:0.03

8:0.05

9:0.09

10:0.05

11:0.02

Negative Logits

��

-1.80

usra

-1.39

edin

-1.35

ohn

-1.28

ONSORED

-1.26

ipl

-1.26

 errone

-1.24

Roz

-1.22

wolves

-1.22

isSpecialOrderable

-1.22

POSITIVE LOGITS

cially

1.46

Delivery

1.42

arenthood

1.38

 fulfilled

1.29

uberty

1.25

hement

1.24

Factory

1.23

aldehyde

1.19

ience

1.19

Required

1.19

Activations Density 0.001%

words related to legal and bureaucratic procedures

No Known Activations