INDEX

Explanations

specific identifiers or references in a formal or technical context, often relating to entities like places, scientific terms, or specific nomenclature

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

Palo Alto, athletic ability

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

Negative Logits

+#+

-0.42

 ſou

-0.37

 senhora

-0.37

 ſer

-0.36

 houſe

-0.35

paravant

-0.35

 Perſ

-0.35

 esclavos

-0.35

Eſ

-0.35

bbons

-0.34

POSITIVE LOGITS

expandindo

0.74

AccessorTable

0.57

 Normdatei

0.56

 Roskov

0.55

MemoryWarning

0.53

Trama

0.52

)_/¯

0.52

ValueStyle

0.52

 propOrder

0.51

 الرياضيه

0.50

Activations Density 0.099%

specific identifiers or references in a formal or technical context, often relating to entities like places, scientific terms, or specific nomenclature

Palo Alto, athletic ability

No Known Activations