INDEX

Explanations

attends to contrastive or explanatory tokens from preceding statements or clauses

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

0:0.06

1:0.07

2:0.07

3:0.10

4:0.11

5:0.05

6:0.38

7:0.12

Negative Logits

setupUi

-0.50

})();

-0.42

<bos>

-0.41

 الحره

-0.40

]")]

-0.40

")->

-0.40

 disambiguazione

-0.39

 يتيمه

-0.39

 .”

-0.39

 AssemblyTitle

-0.38

POSITIVE LOGITS

AndEndTag

0.38

 nemlig

0.31

SerializedSize

0.29

 bulunabilir

0.29

లాలు

0.28

 pangan

0.28

 oprot

0.28

 createSprite

0.28

Answers

0.27

깥

0.27

Activations Density 0.501%

attends to contrastive or explanatory tokens from preceding statements or clauses

No Comments

No Known Activations