INDEX

Explanations

phrases related to current events and activities happening in a specific time frame

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

0:0.06

1:0.02

2:0.24

3:0.05

4:0.24

5:0.10

6:0.04

7:0.03

8:0.04

9:0.05

10:0.06

11:0.03

Negative Logits

 Cosponsors

-1.45

grain

-1.41

bowl

-1.39

BC

-1.36

 Flavoring

-1.35

 Tata

-1.34

 Dwell

-1.31

alam

-1.30

MIN

-1.30

recent

-1.29

POSITIVE LOGITS

 undone

1.27

 accelerated

1.27

 affirmation

1.24

 indictment

1.20

 homosexuals

1.18

ournal

1.18

 swaps

1.18

 swapped

1.17

erers

1.16

 licensee

1.16

Activations Density 0.007%

phrases related to current events and activities happening in a specific time frame

No Known Activations