INDEX

Explanations

attends to tokens associated with specific keywords or characters from sequentially related 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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.23

1:0.23

2:0.16

3:0.11

4:0.07

5:0.03

6:0.07

7:0.06

Negative Logits

onsored

-0.26

-0.25

ETHING

-0.25

 ویکی

-0.24

ÑA

-0.24

 Ком

-0.23

hanna

-0.22

adze

-0.22

ேர

-0.22

asonic

-0.22

POSITIVE LOGITS

+#+#

0.43

 Roskov

0.41

 Seeder

0.36

 propOrder

0.35

ंदीखरीदारी

0.34

EndContext

0.33

 DbType

0.33

 <<<<<<<<<<<<<<

0.32

ISupport

0.32

__':

0.32

Activations Density 1.462%

attends to tokens associated with specific keywords or characters from sequentially related tokens

No Comments

No Known Activations