INDEX

Explanations

attends to function-related terms from corresponding expression-related terms

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

0:0.07

1:0.14

2:0.10

3:0.07

4:0.30

5:0.12

6:0.08

7:0.08

Negative Logits

localScale

-0.21

 process

-0.20

 ład

-0.20

ביוגרפיה

-0.20

 formal

-0.20

uland

-0.20

lobo

-0.20

 blind

-0.20

 again

-0.20

 стро

-0.20

POSITIVE LOGITS

SharedDtor

0.33

">—

0.29

 مشين

0.29

 Stub

0.28

 UIApplication

0.28

 estimés

0.27

 SwitchCompat

0.27

 PopupWindow

0.27

GNP

0.27

 సౌకర్య

0.27

Activations Density 0.230%

attends to function-related terms from corresponding expression-related terms

No Comments

No Known Activations