INDEX

Explanations

attends to numerical tokens from publication details or references from citations

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

0:0.16

1:0.22

2:0.14

3:0.08

4:0.08

5:0.05

6:0.07

7:0.16

Negative Logits

 GenerationType

-0.43

marle

-0.40

 financière

-0.40

Jeografia

-0.40

 كومونز

-0.39

 financières

-0.39

 Acapulco

-0.39

 Normdatei

-0.38

^(@)

-0.38

lawayo

-0.37

POSITIVE LOGITS

cor

0.28

mary

0.27

wts

0.26

Cor

0.26

 King

0.26

êmio

0.25

principalTable

0.25

jar

0.24

sp

0.23

str

0.23

Activations Density 0.035%

attends to numerical tokens from publication details or references from citations

No Comments

No Known Activations