INDEX

Explanations

attends to the token that is marked with brackets from the preceding token marked with asterisks

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

0:0.07

1:0.09

2:0.07

3:0.10

4:0.10

5:0.07

6:0.33

7:0.14

Negative Logits

IsMutable

-0.57

verwijspagina

-0.54

AndEndTag

-0.52

__':

-0.46

__':

-0.44

 betweenstory

-0.43

SharedDtor

-0.42

 "..\..\

-0.40

fromnode

-0.39

 Мексичка

-0.38

POSITIVE LOGITS

اتی

0.31

0.30

ada

0.29

Durée

0.29

awa

0.28

莫

0.28

 college

0.28

 عل

0.28

ins

0.28

Att

0.28

Activations Density 0.176%

attends to the token that is marked with brackets from the preceding token marked with asterisks

No Comments

No Known Activations