INDEX

Explanations

attends to code-related tokens from various programming language constructs

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

0:0.14

1:0.15

2:0.11

3:0.12

4:0.11

5:0.05

6:0.08

7:0.20

Negative Logits

 تضيفلها

-0.35

contentLoaded

-0.33

tagHelperRunner

-0.33

WebVitals

-0.29

Jîn

-0.28

 HFILL

-0.28

RunAsync

-0.28

 Administrativna

-0.27

SourceChecksum

-0.27

Varint

-0.26

POSITIVE LOGITS

исленность

0.31

://$

0.29

InstanceState

0.29

esModule

0.29

setupUi

0.28

 coute

0.28

AttributeSet

0.27

 soyez

0.26

Bukan

0.26

Quiero

0.26

Activations Density 0.001%

attends to code-related tokens from various programming language constructs

No Comments

No Known Activations