INDEX

Explanations

phrases indicating completeness or totality

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

"whole" or "entire"

np_acts-logits-general · gemini-2.0-flash

whole or entire things

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-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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Datuak

-0.49

/*++

-0.45

APN

-0.40

 ARIA

-0.40

dnn

-0.39

 nonUne

-0.39

structors

-0.39

 Epson

-0.39

 diseñadores

-0.38

Zeneca

-0.38

POSITIVE LOGITS

 entire

1.22

whole

1.22

 whole

1.21

entire

1.14

Whole

1.13

 WHOLE

1.13

Entire

1.07

 Whole

1.04

 Entire

1.04

 ENTIRE

1.00

Activations Density 0.074%

phrases indicating completeness or totality

"whole" or "entire"

whole or entire things

No Comments

No Known Activations