INDEX

Explanations

attends to punctuation marks from subsequent tokens

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.10

1:0.11

2:0.12

3:0.16

4:0.09

5:0.03

6:0.18

7:0.16

Negative Logits

ViewFeatures

-0.30

-0.28

ty

-0.27

uk

-0.27

<eos>

-0.27

ric

-0.27

to

-0.26

Portale

-0.26

met

-0.25

 hard

-0.25

POSITIVE LOGITS

 purpoſe

0.40

 ―――――

0.40

 Reſ

0.40

 ་་

0.39

 myſelf

0.39

 houſe

0.38

 Houſe

0.37

/−

0.37

 greateſt

0.36

 itſelf

0.36

Activations Density 0.000%

attends to punctuation marks from subsequent tokens

No Comments

No Known Activations