INDEX

Explanations

attends to functional transformation tokens from accessibility-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

0:0.11

1:0.15

2:0.12

3:0.12

4:0.11

5:0.05

6:0.11

7:0.20

Negative Logits

Personendaten

-0.31

httphttps

-0.30

herme

-0.30

//

-0.29

 CreateTagHelper

-0.29

 initComponents

-0.28

 koszulka

-0.27

FDP

-0.26

 gynhyrchwyd

-0.26

bardier

-0.26

POSITIVE LOGITS

 InputDecoration

0.26

AddField

0.25

RectangleBorder

0.25

PDOException

0.25

showConfirm

0.24

chläge

0.23

 liner

0.23

RSpec

0.23

 noqa

0.23

 الاطلاع

0.23

Activations Density 0.144%

attends to functional transformation tokens from accessibility-related tokens

No Comments

No Known Activations