INDEX

Explanations

Podcast episode

np_max-act · gemini-2.0-flash

New Auto-Interp

Configuration

mwhanna/qwen3-4b-transcoders/layer_22.safetensors

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

163,840

Data Type

float32

Hook Name

blocks.22.mlp.hook_in

Architecture

transcoder

Context Size

8,192

Dataset

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

ìłĲìĿĦ

-0.28

 treatment

-0.27

xit

-0.25

brand

-0.24

ln

-0.24

åĨĻéģĵ

-0.24

çµ

-0.24

å¸ĸ

-0.24

è±¸

-0.24

bons

-0.23

POSITIVE LOGITS

/audio

0.29

å½ķåĪ¶

0.28

.wav

0.26

DidLoad

0.25

æĴŃæĶ¾

0.25

rouch

0.25

 TRAN

0.25

æĴŃåĩº

0.25

_TRAN

0.25

pÃ©ri

0.24

Activations Density 0.010%

Podcast episode

No Known Activations