INDEX

Explanations

references to numerical data related to draft picks and financial commitments

New Auto-Interp

Top Features by Cosine Similarity

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

2:0.16

3:0.06

4:0.13

5:0.05

6:0.06

7:0.05

8:0.07

9:0.06

10:0.12

11:0.16

Negative Logits

akery

-1.64

�醒

-1.48

aptic

-1.47

eteenth

-1.39

hetti

-1.33

asser

-1.31

atre

-1.30

obyl

-1.29

andestine

-1.28

��

-1.27

POSITIVE LOGITS

 apiece

1.61

 totaling

1.60

 evenly

1.39

ヴァ

1.33

 total

1.32

 proportions

1.28

 sizes

1.27

 consisting

1.27

TPPStreamerBot

1.24

 averaging

1.23

Activations Density 0.101%

references to numerical data related to draft picks and financial commitments

No Known Activations