INDEX

Explanations

attends to various tokens from previous mentions of the same named entity or concept

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

0:0.04

1:0.09

2:0.05

3:0.05

4:0.17

5:0.47

6:0.05

7:0.04

Negative Logits

躇

-0.38

 BoxFit

-0.37

fjspx

-0.35

invokeLater

-0.35

Havolalar

-0.34

 WebDriverWait

-0.34

 Quim

-0.34

 تضيفلها

-0.33

setVerticalGroup

-0.33

énario

-0.32

POSITIVE LOGITS

</em>

0.38

`${

0.28

AutoScale

0.27

mnop

0.27

 يتيمه

0.27

 ***/

0.26

 Locust

0.26

 contará

0.25

ceptre

0.25

úgó

0.25

Activations Density 0.101%

attends to various tokens from previous mentions of the same named entity or concept

No Comments

No Known Activations