INDEX

Explanations

occurrences of the word 'des' indicating quantities or quantities related to items

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

des + nouns

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

 Wikiseite

-0.61

 héri

-0.53

RegressionTest

-0.53

 enfans

-0.52

 vainqueur

-0.52

 honneur

-0.51

 ennemi

-0.50

 keja

-0.49

 âme

-0.49

 chrétien

-0.48

POSITIVE LOGITS

 different

0.78

 some

0.68

 small

0.67

an

0.66

 special

0.64

 bigger

0.64

0.63

 great

0.63

few

0.63

 dozen

0.61

Activations Density 0.004%

occurrences of the word 'des' indicating quantities or quantities related to items

des + nouns

No Comments

No Known Activations