INDEX

Explanations

phrases indicating exclusivity or uniqueness

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

0:0.07

1:0.08

2:0.08

3:0.08

4:0.09

5:0.08

6:0.08

7:0.07

8:0.08

9:0.07

10:0.07

11:0.08

Negative Logits

Alert

-2.72

Sov

-2.72

�

-2.71

Camp

-2.56

ARM

-2.49

ttle

-2.42

Ele

-2.37

 Dare

-2.35

Shock

-2.35

defense

-2.33

POSITIVE LOGITS

 Gibraltar

2.67

Quote

2.52

 parted

2.52

 reconnect

2.38

 alignment

2.36

 Chips

2.35

raltar

2.31

 Thatcher

2.31

 Beckham

2.27

 cartel

2.26

Activations Density 0.000%

phrases indicating exclusivity or uniqueness

No Known Activations