INDEX

Explanations

attends to the token "patterns" from empty tokens or brackets

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

0:0.13

1:0.16

2:0.10

3:0.10

4:0.14

5:0.16

6:0.07

7:0.09

Negative Logits

 MonoBehaviour

-0.31

CONSIN

-0.30

brainly

-0.28

 Infórmanos

-0.28

NUMX

-0.27

astify

-0.27

Datuak

-0.26

kuuta

-0.26

IContainer

-0.25

colhead

-0.24

POSITIVE LOGITS

 typelib

0.29

 ویکی‌پدیا

0.28

vitis

0.25

 surla

0.25

Scy

0.25

 صوتيه

0.25

 oprot

0.25

 années

0.25

inali

0.25

BeanUtils

0.25

Activations Density 0.004%

attends to the token "patterns" from empty tokens or brackets

No Comments

No Known Activations