INDEX

Explanations

attends to category labels marked as categories, such as numerical identifiers and types of entities, from their respective references or classifications

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.08

1:0.13

2:0.08

3:0.06

4:0.05

5:0.02

6:0.19

7:0.36

Negative Logits

 autorytatywna

-0.34

ExtendWith

-0.33

board

-0.32

ReusableCell

-0.30

mybatisplus

-0.30

 BoxDecoration

-0.29

EndInit

-0.28

BOARD

-0.27

 InputDecoration

-0.26

Board

-0.26

POSITIVE LOGITS

SourceChecksum

0.42

RegistryLite

0.32

httphttps

0.30

 amongst

0.29

 écl

0.29

ixon

0.28

leyebilirsiniz

0.28

itorious

0.28

Geplaatst

0.28

 volontà

0.28

Activations Density 0.144%

attends to category labels marked as categories, such as numerical identifiers and types of entities, from their respective references or classifications

No Comments

No Known Activations