INDEX

Explanations

attends to tokens relating to the concepts of eating or authority from tokens indicating opposing concepts or entities

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

1:0.10

2:0.08

3:0.09

4:0.11

5:0.03

6:0.26

7:0.22

Negative Logits

});

-0.49

 Bertie

-0.47

#%%

-0.46

',

-0.45

")));

-0.45

)_/¯

-0.44

 Normdatei

-0.44

%%

-0.44

 Sikhs

-0.43

.'</

-0.43

POSITIVE LOGITS

Abitanti

0.40

0.38

【

0.36

BrowserModule

0.34

bitField

0.34

all

0.33

gand

0.33

and

0.33

djangoproject

0.33

Activations Density 0.061%

attends to tokens relating to the concepts of eating or authority from tokens indicating opposing concepts or entities

No Comments

No Known Activations