INDEX

Explanations

Chinese text

np_max-act-logits · gemini-2.0-flash

New Auto-Interp

Configuration

mwhanna/qwen3-4b-transcoders/layer_30.safetensors

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

163,840

Data Type

float32

Hook Name

blocks.30.mlp.hook_in

Architecture

transcoder

Context Size

8,192

Dataset

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 China

-0.35

 Tian

-0.34

 Shanghai

-0.32

 Mandarin

-0.32

 Zhao

-0.30

China

-0.30

ä¸ŃåĽ½å¤§éĻĨ

-0.29

 Beijing

-0.29

å¸¦æĿ¥

-0.29

çī¯

-0.28

POSITIVE LOGITS

allon

0.27

emy

0.27

rug

0.27

rays

0.26

åľĨæ»¡

0.24

asm

0.24

untas

0.24

EU

0.24

 bundles

0.23

 lact

0.23

Activations Density 0.008%

Chinese text

No Known Activations