INDEX

Explanations

attends to a variety of tokens marked with specific identifiers from tokens also marked with specific identifiers

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

1:0.07

2:0.06

3:0.04

4:0.11

5:0.55

6:0.05

7:0.04

Negative Logits

出版年

-0.29

 Divulgação

-0.24

CompleteListener

-0.23

 Thurman

-0.22

gameserver

-0.21

YLON

-0.21

Geplaatst

-0.21

annotate

-0.21

 fumée

-0.21

ripsi

-0.21

POSITIVE LOGITS

+#+#

0.33

UrlResolution

0.31

])));

0.27

 combineReducers

0.26

."));

0.26

EndProject

0.26

abetes

0.26

setAccessible

0.25

RectangleBorder

0.25

ğiniz

0.25

Activations Density 0.176%

attends to a variety of tokens marked with specific identifiers from tokens also marked with specific identifiers

No Comments

No Known Activations