INDEX

Explanations

attends to the last image token from image-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

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.10

1:0.13

2:0.06

3:0.06

4:0.13

5:0.31

6:0.05

7:0.12

Negative Logits

 ALLA

-0.31

agot

-0.31

liferay

-0.31

теше

-0.30

ilanth

-0.30

بالإ

-0.30

GrantedAuthority

-0.30

 Betracht

-0.30

 além

-0.30

 Hawthorne

-0.30

POSITIVE LOGITS

harusnya

0.39

 تانيه

0.30

ddots

0.30

 okuyayım

0.30

 sumpay

0.29

τως

0.28

classnames

0.28

xffffffff

0.28

ImageContext

0.27

Activations Density 0.009%

attends to the last image token from image-related tokens

No Comments

No Known Activations