INDEX

Explanations

attends to tokens that follow numerical indicators or lists from earlier 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

0:0.08

1:0.15

2:0.45

3:0.08

4:0.06

5:0.03

6:0.04

7:0.07

Negative Logits

pixels

-0.32

SWR

-0.30

didSet

-0.30

 Halen

-0.30

>//

-0.30

heights

-0.29

surfaces

-0.29

Ownership

-0.28

snip

-0.28

ienki

-0.28

POSITIVE LOGITS

 HasFactory

0.48

httphttps

0.47

CodeDom

0.42

makeConstraints

0.42

 ویکی‌پدیا

0.41

 følge

0.41

 inoxydable

0.40

tilhe

0.40

 DbType

0.40

LookAnd

0.39

Activations Density 0.277%

attends to tokens that follow numerical indicators or lists from earlier tokens

No Comments

No Known Activations