INDEX

Explanations

Tokyo and Korea

np_max-act · gemini-2.0-flash

New Auto-Interp

Configuration

mwhanna/qwen3-4b-transcoders/layer_2.safetensors

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

163,840

Data Type

float32

Hook Name

blocks.2.mlp.hook_in

Architecture

transcoder

Context Size

8,192

Dataset

monology/pile-uncopyrighted

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 wrong

-0.28

Ð¿Ð¸ÑĤ

-0.26

ses

-0.26

æģ£

-0.25

.Gr

-0.25

 mist

-0.24

sm

-0.24

 timing

-0.24

 eyeb

-0.24

æľīçĶ¨

-0.23

POSITIVE LOGITS

IGNED

0.26

åħ³åĪĩ

0.25

named

0.24

HIP

0.24

-ln

0.24

ppo

0.24

#:

0.24

istani

0.23

acas

0.23

 contiguous

0.23

Activations Density 36.315%

Tokyo and Korea

No Comments

No Known Activations