INDEX

Explanations

attends to descriptors or identifiers of individuals from tokens expressing qualities or actions related to those individuals

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

0:0.15

1:0.19

2:0.13

3:0.09

4:0.10

5:0.03

6:0.08

7:0.19

Negative Logits

 fubject

-0.39

 houſe

-0.39

 Jefus

-0.38

 fhort

-0.38

IUrlHelper

-0.37

 purpoſe

-0.37

"):

-0.37

 myſelf

-0.37

 Chriftian

-0.37

 ſtate

-0.36

POSITIVE LOGITS

ly

0.30

 Cinque

0.28

 Trek

0.28

TabIndex

0.28

bal

0.27

mena

0.27

0.25

gil

0.25

/***/

0.24

Activations Density 0.308%

attends to descriptors or identifiers of individuals from tokens expressing qualities or actions related to those individuals

No Comments

No Known Activations