INDEX

Explanations

attends to specific keywords from general tokens

oai_attention-head · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 10-gemmascope-att-16k

Configuration

google/gemma-scope-2b-pt-att/layer_10/width_16k/average_l0_70

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.10.attn.hook_z

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.24

1:0.20

2:0.11

3:0.14

4:0.08

5:0.03

6:0.07

7:0.10

Negative Logits

//

-0.36

 Parigi

-0.34

SCS

-0.33

 sond

-0.33

الإنجليزية

-0.33

存于互联网档案馆

-0.32

annica

-0.32

 mergeFrom

-0.32

ък

-0.32

eaway

-0.32

POSITIVE LOGITS

 HasFactory

0.32

Piv

0.31

jupiter

0.30

AccessorTable

0.29

ورده

0.28

 uska

0.28

 Bullock

0.28

 protoimpl

0.27

 VIAF

0.26

weightedMode

0.26

Activations Density 0.312%

attends to specific keywords from general tokens

No Comments

No Known Activations