INDEX

Explanations

attends to mathematical symbols or notations from any associated context or formula

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

0:0.13

1:0.23

2:0.13

3:0.11

4:0.13

5:0.12

6:0.05

7:0.07

Negative Logits

 Sampson

-0.29

 szóci

-0.26

ccione

-0.26

vim

-0.25

 būtų

-0.25

 motivated

-0.24

)});

-0.24

}*/

-0.24

__(/*!

-0.24

}))

-0.23

POSITIVE LOGITS

AutoScaleMode

0.48

 AssemblyCulture

0.46

 Roskov

0.43

principalColumn

0.42

parsedMessage

0.42

 cherchés

0.41

WaitGroup

0.40

ніципалі

0.39

OGND

0.39

хьтан

0.39

Activations Density 0.114%

attends to mathematical symbols or notations from any associated context or formula

No Comments

No Known Activations