INDEX

Explanations

blog posts and comments

np_max-act-logits · gemini-2.0-flash

New Auto-Interp

Configuration

mwhanna/qwen3-4b-transcoders/layer_31.safetensors

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

163,840

Data Type

float32

Hook Name

blocks.31.mlp.hook_in

Architecture

transcoder

Context Size

8,192

Dataset

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

WithType

-0.29

rap

-0.28

 learnt

-0.27

 Learn

-0.25

Mus

-0.25

æĥĬ

-0.24

 fern

-0.24

 ingress

-0.24

åħ¥ä½ı

-0.23

Hwy

-0.23

POSITIVE LOGITS

 then

0.33

ç»ĵ

0.28

à¹ģà¸¥

0.28

then

0.28

åĨį

0.28

æīĢæľīäºº

0.27

 Then

0.27

Then

0.25

hog

0.24

_then

0.24

Activations Density 0.004%

blog posts and comments

No Known Activations