AMUSE3.1 Shows us NSFW Images!! (written English)
Notice: This article is a rewrite of this content in English.
And part of it was rewitten.
Hello, AMUSERs!! (AMUSE + USERs)
It may be a little unsightly English, but please forgive me.
Immediately, as written in the headline, I will explain how to output sensitive pictures with "AMUSE3.1 a local generation AI tool" !!
…It's not 2.x.
about 3.1. (The current version at the time of writing)
General procedure
Installation of AMUSE v3.1.0
Install onnx2json
Backup Amuse\Plugins\ContentFilter\ContentFilter.onnx to any folder
Duplicate and edit the backup ContentFilter.onnx file
Start Amuse.exe
I'm actually going to give NSFW instructions
Dissatisfied
I'm done
Let's take a look.
By theway, this author's PC configuration is as follows, I hope it will be helpful.
--you can enjoy generation even on a computer of this level--
OS: Windows10 Home
CPU: Ryzen5 2600 (Gen_zen1+)
RAM: 32GByte (about 16GB is dangerous…)
GPU: Radeon RX6700XT 12GB
Installation of AMUSE v3.1.0
<< I will omit a detailed explanation!! >>
Download this installation file and install it anywhere. It's easy to use if you have about 200GB of free space.
Install onnx2json
Let's ask AI at a time like this. (copilot, chat-GPT,or others…)
Download and run python-3.13.x-amd64.exe .
(The Python version should be the latest.)
When you are done, open the command prompt…
pip install -U onnx2jsonHit and onnx2json will be installed.
json2onnx is also included here, so let's install it.
pip install -U onnx protobuf
pip install -U json2onnxIt gets complicated, so the command prompt is finished once.
Backup Amuse\Plugins\ContentFilter\ContentFilter.onnx to any folder
Let evacuate ContentFilter.onnx for the time being.
Copy the file to Amuse\Backup as an example.
( I placed it directly below Disk E: )
Duplicate and edit the backup ContentFilter.onnx file
Duplicate the ContentFilter.onnx and Rename it to ContentFilterMod.onnx.
Continue to open the command prompt in this folder.
When the command prompt starts,
onnx2json --input_onnx_file_path ContentFilterMod.onnx --output_json_path ContentFilterMod.jsonand enter. It will takes a while (about 30sec…)
There are no messages in particular, but ContentFilterMod.json is created in the folder.
Edit this ContentFilterMod.json.
This time you will edit it with Notepad++ ( Can you use simpley Notepad? )
ここから先は
この記事が気に入ったらチップで応援してみませんか?


購入者のコメント
4試してみましたが残念ながらできませんでした。
◆version
Amuse v3.1.0
◆エラーコード
[ErrorCode:Fail] Deserialize tensor
concept_embeds_weights failed.endian_utils.cc:57
onnxruntime::utils::detail::CopyLittleEndian source and
destination buffer size mismatch
◆jsonファイル変更箇所
※指定の場所を変更
ゲストさん
ご報告ありがとうございます。
原因が分かり次第コメントいたします。
【解決済】↑のコメントをしたゲストです。
購入後日本語に翻訳したから駄目だったようです。
英語のままの記事の文字列と翻訳後で勝手に文字列が返還されて差し込まれてしまっていました。ご迷惑をおかけして申し訳ありません。現在できるようになってます。
ゲストさん
ご報告いただきまして誠にありがとうございます。
無事解決なされたようで何よりです!