INDEX

Explanations

articles and prepositions in the text

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

names of places or buildings

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

 aplicable

-0.43

 consists

-0.43

collana

-0.42

 consiste

-0.41

essentiel

-0.40

 intensidad

-0.40

 encaje

-0.39

 consisted

-0.39

 continuidad

-0.39

coration

-0.38

POSITIVE LOGITS

 مرئيه

0.65

 barracks

0.61

ViewChild

0.59

 courthouse

0.59

 hotel

0.56

ValueStyle

0.56

 clubhouse

0.55

 synagogue

0.55

 Hotel

0.54

 mosque

0.54

Activations Density 0.010%

articles and prepositions in the text

names of places or buildings

No Comments

No Known Activations