INDEX

Explanations

terms associated with common denominators and commonality in concepts or entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Staates

-0.60

 Turquía

-0.57

 GenerationType

-0.56

 Aéroport

-0.54

TypedDataSet

-0.54

iermark

-0.51

 Inscrivez

-0.51

stateProvider

-0.48

igurumi

-0.48

 Irán

-0.47

POSITIVE LOGITS

 common

1.22

 Common

1.22

Common

1.20

common

1.18

 COMMON

1.03

COMMON

0.93

 CommonModule

0.91

CommonModule

0.89

 común

0.87

 commons

0.85

Activations Density 0.105%

terms associated with common denominators and commonality in concepts or entities

No Comments

No Known Activations