INDEX

Explanations

attends to prepositions or linking phrases from subsequent tokens indicating specific ownership or association

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

0:0.10

1:0.14

2:0.13

3:0.10

4:0.10

5:0.03

6:0.12

7:0.22

Negative Logits

Vorlage

-0.26

aways

-0.26

Dichloroethane

-0.26

ďaka

-0.26

UrlResolution

-0.25

 famí

-0.25

########.

-0.24

ableness

-0.24

Климат

-0.24

overs

-0.24

POSITIVE LOGITS

ExtendWith

0.30

expandindo

0.29

ScopeManager

0.27

LEncoder

0.26

 ویکی‌پدیای

0.26

InSection

0.25

awaiter

0.25

ValueStyle

0.25

 onPostExecute

0.24

 MainAxisSize

0.24

Activations Density 0.050%

attends to prepositions or linking phrases from subsequent tokens indicating specific ownership or association

No Comments

No Known Activations