INDEX

Explanations

attends to tokens related to scores and classifications from numeric 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

0:0.04

1:0.08

2:0.04

3:0.04

4:0.20

5:0.51

6:0.03

7:0.03

Negative Logits

<?

-0.32

NOPQRST

-0.31

ToProps

-0.29

󠁮

-0.27

}:

-0.26

 RIPRODUZIONE

-0.25

DropColumn

-0.25

ReusableCell

-0.24

Pozri

-0.24

AndEndTag

-0.24

POSITIVE LOGITS

gensen

0.35

titleMargin

0.35

 Италијани

0.35

opha

0.35

MessageOf

0.34

횟

0.34

ValueStyle

0.34

rafa

0.33

 الدولى

0.32

rungsseite

0.32

Activations Density 0.329%

attends to tokens related to scores and classifications from numeric tokens

No Comments

No Known Activations