INDEX

Explanations

attends to class-related tokens from corresponding attribute 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

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.18

1:0.13

2:0.06

3:0.07

4:0.07

5:0.12

6:0.05

7:0.29

Negative Logits

 архивлан

-0.45

+#+#

-0.40

GEBURTS

-0.37

PreferredItem

-0.37

 centralised

-0.37

SequentialGroup

-0.37

 organisers

-0.35

 skilful

-0.35

otides

-0.35

 Cunha

-0.35

POSITIVE LOGITS

 ferons

0.36

行

0.32

 חיצוניים

0.30

feira

0.28

 représ

0.28

亚

0.28

complish

0.28

Спасылкі

0.28

St

0.27

jest

0.27

Activations Density 0.001%

attends to class-related tokens from corresponding attribute tokens

No Comments

No Known Activations