INDEX

Explanations

attends to tokens marked as "this" from any tokens marked with "that."

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

0:0.08

1:0.12

2:0.49

3:0.07

4:0.05

5:0.02

6:0.04

7:0.07

Negative Logits

bibinfo

-0.38

aarrggbb

-0.38

rawDesc

-0.37

pushFollow

-0.35

 doubtnut

-0.34

addContainerGap

-0.34

glades

-0.34

˟

-0.33

onViewCreated

-0.33

BibitemOpen

-0.33

POSITIVE LOGITS

argout

0.29

0.28

انيف

0.28

 Quelles

0.27

IsContent

0.27

”

0.26

homonymie

0.26

formik

0.26

 appunto

0.26

Activations Density 0.552%

attends to tokens marked as "this" from any tokens marked with "that."

No Comments

No Known Activations