INDEX

Explanations

attends to tokens that indicate additional information or context from subsequent tokens that also provide additional details

oai_attention-head · gpt-4o-mini Triggered by @bot

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-att/layer_0/width_16k/average_l0_61

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.0.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

0:0.01

1:0.00

2:0.02

3:0.01

4:0.00

5:0.01

6:0.61

7:0.10

8:0.00

9:0.01

10:0.00

11:0.03

12:0.01

13:0.08

14:0.02

15:0.01

Negative Logits

 autorytatywna

-0.99

 ویکی‌پدی

-0.87

uxxxx

-0.75

 Италијани

-0.73

 noDo

-0.68

IUrlHelper

-0.68

 betweenstory

-0.68

ंदीखरीदारी

-0.68

 Мексичка

-0.65

 kaarangay

-0.64

POSITIVE LOGITS

0.27

0.23

0.22

↵

0.21

0.20

Activations Density 0.160%

attends to tokens that indicate additional information or context from subsequent tokens that also provide additional details

No Comments

No Known Activations