INDEX

Explanations

a pattern not present in these excerpts; it remains inactive and likely targets a rare or absent feature.

oai_token-act-pair · gpt-5 Triggered by @liuliujiujiu05

New Auto-Interp

Configuration

mwhanna/qwen3-4b-transcoders/layer_1.safetensors

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

163,840

Data Type

float32

Hook Name

blocks.1.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

iminal

-0.29

è´°

-0.28

skl

-0.27

Dou

-0.27

æĹ¥æŃ£å¼ı

-0.26

fat

-0.25

 Spir

-0.25

 asset

-0.25

 inertia

-0.24

asset

-0.24

POSITIVE LOGITS

çľĭä»ĸ

0.27

wahl

0.27

iform

0.25

builders

0.25

 soon

0.25

 hurry

0.25

nah

0.24

æĮ¯åĬ¨

0.24

ä¸įå·²

0.24

ä¸įä¹ħ

0.24

Activations Density 21.937%

No Known Activations

This feature has no known activations.