INDEX

Explanations

attends to mathematical or comparative operators from tokens that express comparisons or categories

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

0:0.11

1:0.13

2:0.10

3:0.12

4:0.12

5:0.04

6:0.17

7:0.16

Negative Logits

 alamat

-0.25

 ehre

-0.22

</h6>

-0.22

 rispetto

-0.22

декс

-0.22

 umani

-0.20

</h2>

-0.20

 spät

-0.20

misu

-0.20

reactstrap

-0.20

POSITIVE LOGITS

最快更新

0.38

Diweddarwch

0.37

 ddelweddau

0.34

ReusableCell

0.34

 externi

0.34

 незавершена

0.33

RefNanny

0.31

verwijspagina

0.31

RTEE

0.30

########.

0.30

Activations Density 0.000%

attends to mathematical or comparative operators from tokens that express comparisons or categories

No Comments

No Known Activations