INDEX

Explanations

attends to the token "parts" from various segments of text that follow the attention pattern with congruent fillings or descriptions

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

0:0.00

1:0.01

2:0.62

3:0.02

4:0.00

5:0.00

6:0.02

7:0.03

8:0.00

9:0.00

10:0.00

11:0.00

12:0.00

13:0.09

14:0.09

15:0.00

Negative Logits

of

-0.40

 word

-0.35

one

-0.35

 world

-0.34

 domain

-0.34

 comp

-0.33

age

-0.33

 results

-0.33

bug

-0.32

 board

-0.32

POSITIVE LOGITS

 mijne

0.31

 resourceCulture

0.30

 anún

0.27

 którzy

0.27

 goederen

0.27

 zijne

0.26

 vertellen

0.26

 onderdelen

0.25

 suaminya

0.25

 allemaal

0.25

Activations Density 0.048%

attends to the token "parts" from various segments of text that follow the attention pattern with congruent fillings or descriptions

No Comments

No Known Activations