INDEX

Explanations

attends to tokens marked as subscripts from tokens also marked as subscripts

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

0:0.05

1:0.08

2:0.05

3:0.04

4:0.23

5:0.44

6:0.03

7:0.05

Negative Logits

 مرئيه

-0.47

aarrggbb

-0.42

TypedDataSet

-0.39

uxxxx

-0.37

addCriterion

-0.35

SourceChecksum

-0.35

mstyle

-0.34

/*

-0.34

+#+#

-0.33

/**

-0.32

POSITIVE LOGITS

 esterni

0.23

ські

0.21

 государственной

0.20

 estekak

0.20

gelöst

0.20

er

0.20

 assai

0.20

 Vaid

0.20

UnusedPrivate

0.20

ローチ

0.20

Activations Density 0.593%

attends to tokens marked as subscripts from tokens also marked as subscripts

No Comments

No Known Activations