INDEX

Explanations

phrases related to product features and safety

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

compared to, complies with, can integrate

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

parsedMessage

-0.74

 propOrder

-0.72

DockStyle

-0.67

 Италијани

-0.57

➯

-0.55

GEBURTSDATUM

-0.54

MessageOf

-0.54

цездатний

-0.53

ToAction

-0.52

TagMode

-0.51

POSITIVE LOGITS

iedades

0.35

 properties

0.34

 eficiencia

0.33

properties

0.33

 Parcelable

0.32

 durability

0.32

engu

0.31

кож

0.30

guar

0.29

 FEATURES

0.29

Activations Density 0.005%

phrases related to product features and safety

compared to, complies with, can integrate

No Comments

No Known Activations