INDEX

Explanations

attends to tokens related to types of writing processes or editing roles from accompanying tokens that appear later in the sentence

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

1:0.17

2:0.07

3:0.10

4:0.15

5:0.24

6:0.08

7:0.07

Negative Logits

 Infórmanos

-0.50

 disambiguazione

-0.49

AsUp

-0.47

RegressionTest

-0.47

saraba

-0.46

новништво

-0.46

 typelib

-0.46

IANGLES

-0.45

يكب

-0.43

 ujednoznacz

-0.43

POSITIVE LOGITS

if

0.32

for

0.28

est

0.27

in

0.26

ət

0.26

at

0.26

ätie

0.25

 much

0.25

Activations Density 0.005%

attends to tokens related to types of writing processes or editing roles from accompanying tokens that appear later in the sentence

No Comments

No Known Activations