INDEX

Explanations

terms and expressions that indicate legal or formal discourse

oai_token-act-pair · gpt-4o-mini Triggered by @bot

common phrase completions

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Penga

-0.36

 juſ

-0.34

 centrales

-0.33

 industriel

-0.32

 pañ

-0.30

 peuple

-0.30

 electrico

-0.30

 preescolar

-0.29

 Lingkungan

-0.29

 reven

-0.29

POSITIVE LOGITS

PerformLayout

0.73

AsUp

0.68

WebElementEntity

0.65

 otomatig

0.63

classnames

0.59

LookAnd

0.58

abstractmethod

0.58

 مرئيه

0.57

 typelib

0.56

TemporalType

0.55

Activations Density 0.447%

terms and expressions that indicate legal or formal discourse

common phrase completions

No Comments

No Known Activations