INDEX

Explanations

instances of the article "a" and variations of "us."

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

pronoun followed by action or state

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

 quæ

-0.37

ReusableCell

-0.37

 entendido

-0.31

álně

-0.31

 estamp

-0.30

 dessutom

-0.30

findpost

-0.30

 promocional

-0.30

Werdegang

-0.30

ьажоргаш

-0.30

POSITIVE LOGITS

nloa

0.56

="@+

0.52

utives

0.52



0.51

 Wicidata

0.49

 okuyayım

0.48

+#+#

0.48

 Violent

0.47

issory

0.47

devamını

0.47

Activations Density 0.001%

instances of the article "a" and variations of "us."

pronoun followed by action or state

No Comments

No Known Activations