INDEX

Explanations

highly emphasized concepts and/or values related to health or wellness

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

Appellant

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

ftagPool

-0.73

ModelSerializer

-0.63

 FetchType

-0.62

 Wikimedijinoj

-0.60

Datuak

-0.59

SBATCH

-0.54

RectangleBorder

-0.54

galus

-0.54

 متعلقه

-0.53

OCCURRED

-0.53

POSITIVE LOGITS

 JButton

0.35

 exib

0.33

Part

0.32

Mr

0.32

fragment

0.32

parallel

0.31

 Plummer

0.31

 memorandum

0.30

öl

0.30

part

0.30

Activations Density 0.001%

highly emphasized concepts and/or values related to health or wellness

Appellant

No Comments

No Known Activations