INDEX

Explanations

attends to tokens indicating planned or scheduled events or actions from tokens specifying the type or nature of those events or actions

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

0:0.10

1:0.15

2:0.10

3:0.18

4:0.13

5:0.05

6:0.09

7:0.15

Negative Logits

enumii

-0.28

 Hodges

-0.25

impulan

-0.25

ACTED

-0.25

metingen

-0.25

melden

-0.24

 répé

-0.24

 consac

-0.23

idamente

-0.23

dasarkan

-0.23

POSITIVE LOGITS

دانشنامهٔ

0.36

//

0.30

 SwitchCompat

0.29

кипедия

0.29

BufferException

0.27

<&

0.27

yntaxException

0.27

urtstag

0.27

ArgsConstructor

0.26

Personensuche

0.26

Activations Density 0.004%

attends to tokens indicating planned or scheduled events or actions from tokens specifying the type or nature of those events or actions

No Comments

No Known Activations