INDEX

Explanations

phrases related to trials and experiences with products

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

try, test, display

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

 Administrativna

-0.50

fromnode

-0.49

จบ

-0.44

CppMethod

-0.43

 ujednoznacz

-0.43

 Comunicación

-0.41

 desmotivaciones

-0.40

 igång

-0.40

對象

-0.40

 Propiedad

-0.39

POSITIVE LOGITS

featureID

0.49

Hw

0.46

kin

0.45

 Ultima

0.45

 transfieras

0.45

oa̍t

0.45

 estekak

0.44

 kits

0.44

Sn

0.44

trial

0.44

Activations Density 0.010%

phrases related to trials and experiences with products

try, test, display

No Comments

No Known Activations