INDEX

Explanations

There isn’t enough evidence to infer its function, since it never activates on any tokens in the provided samples.

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

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

ä¸¤ä½į

-0.27

éĮ¾

-0.26

scatter

-0.26

å°±è¿ŀ

-0.25

è¿Ļæł·çļĦ

-0.25

 scatter

-0.24

 apprent

-0.24

ä¸īä½į

-0.23

å°±è¯´

-0.23

 apprec

-0.23

POSITIVE LOGITS

imed

0.33

posted

0.28

 shells

0.26

ç´¯ç´¯

0.25

orias

0.25

 higher

0.25

uran

0.25

é«ĺä½İ

0.24

eping

0.24

atorial

0.24

Activations Density 38.583%

No Known Activations

This feature has no known activations.