INDEX

Explanations

attends to action or process-related tokens from related subject tokens

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

0:0.14

1:0.15

2:0.17

3:0.07

4:0.06

5:0.04

6:0.09

7:0.25

Negative Logits

 eccell

-0.32

 وتسجيلات

-0.31

 vertes

-0.30

 sauvages

-0.29

xtext

-0.26

Петер

-0.26

 porcelaine

-0.25

 featured

-0.25

 regardless

-0.25

 attended

-0.25

POSITIVE LOGITS

 незавершена

0.38

RegistryLite

0.37

 utafitiHapana

0.37

 '\\;'

0.33

 препратки

0.33

ertale

0.33

клопе

0.32



0.31

AndroidJUnit

0.30

ナメント

0.30

Activations Density 0.764%

attends to action or process-related tokens from related subject tokens

No Comments

No Known Activations