INDEX

Explanations

attends to tokens with specific characters from associated tokens marked by the square bracket

oai_attention-head · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 0-gemmascope-att-16k

Configuration

google/gemma-scope-2b-pt-att/layer_0/width_16k/average_l0_104

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.0.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

0:0.80

1:0.00

2:0.01

3:0.01

4:0.09

5:0.02

6:0.01

7:0.02

Negative Logits

Amicalement

-0.44

BibitemShut

-0.42

Efq

-0.38

 Majefty

-0.38

 PicClick

-0.37

ArrowToggle

-0.36

 myſelf

-0.35

]--;

-0.35

bibfield

-0.35

$_"

-0.35

POSITIVE LOGITS

↵↵

0.40

So

0.30

The

0.30

 When

0.28

↵

0.28

0.27

Any

0.26

󠁢

0.26

↵↵↵

0.26

 This

0.25

Activations Density 5.918%

attends to tokens with specific characters from associated tokens marked by the square bracket

No Comments

No Known Activations