INDEX

Explanations

attends to method-related tokens from corresponding tokens related to its application or result

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

0:0.09

1:0.11

2:0.10

3:0.11

4:0.09

5:0.04

6:0.17

7:0.24

Negative Logits

Adv

-0.26

 Avan

-0.25

ous

-0.25

latego

-0.24

chol

-0.24

*-*-

-0.24

 Beet

-0.24

tum

-0.24

луб

-0.23

wpi

-0.23

POSITIVE LOGITS

LookAnd

0.35

 CreateTagHelper

0.34

transQ

0.32

 StringTokenizer

0.31

__))

0.31

 فريبيس

0.31

adpleegd

0.31

IVEREF

0.31

illère

0.31

}))

0.31

Activations Density 0.329%

attends to method-related tokens from corresponding tokens related to its application or result

No Comments

No Known Activations