INDEX

Explanations

attends to the second token in the sequence from earlier tokens in the same sequence

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

0:0.10

1:0.15

2:0.37

3:0.09

4:0.07

5:0.02

6:0.06

7:0.11

Negative Logits

 للمعارف

-0.34

SourceChecksum

-0.31

enumi

-0.31

featureID

-0.30

CreateModel

-0.30

脚注の使い方

-0.29

ariConfig

-0.29

 المعيارى

-0.28

sidemargin

-0.28

 يتيمه

-0.27

POSITIVE LOGITS

 }}"></

0.31

ühner

0.28

 Mühl

0.28

 SwiftUI

0.28

 Whoever

0.27

endió

0.26

 chrétienne

0.26

 Prudential

0.26

 kini

0.26

Gud

0.25

Activations Density 1.082%

attends to the second token in the sequence from earlier tokens in the same sequence

No Comments

No Known Activations