obsidian-SenGener icon indicating copy to clipboard operation
obsidian-SenGener copied to clipboard

This plugin is used to generate a serial of Sentenses for writting.

SenGenerPlugin for obsidian 中文

This plugin is used to generate a serial of Sentenses for writting.

Thanks

Sengener based on obsidian-completr. I copied lots of code from it. https://github.com/tth05/obsidian-completr

How to use this plugin

  1. download and install obsidian. Create a vault and specify the path.

  2. goto the vault folder. download from releases , and extract them to the plugin folder, the path like: VAULT/.obsidian/plugins/obsidian-sengener.

  3. Open setting, enable community plugins, then set you options, enable SenSener and configure hot-keys. The default shortcut key is ctrl+quotation.

  4. You can also select different authoring models and adjust other parameters.

  5. Create a document and start writing. Enjoy it. And contact me: [email protected].

Parameter

  • API address: service address : Fill in your own API address. I built an example service, which includes English model, dialogue model(Chinese), work report model(Chinese) and Tencent welm model(Chinese and few English suport). Sample address: https://fwzd.myfawu.com
  • Type: You can select different authoring models.
  • Token: the token used to log in to Tencent welm. You can apply by yourself on Tencent welm official website.
  • Enable searching: Whether to enable full-text retrieval. Currently, it provides full-text retrieval for the report model.
  • Number of choices: How many candidates. Don’t choose too many, which will affect the speed.
  • Max length: the number of words produced at a time. Don’t choose too many, which will affect the speed.

Build your data service

  • I build the API service with GPT2. You can also use GPT2 to generate Sentenses directly , or Another one, like yuanyu, or chatGPT.

  • You can build your API service for generating Sentenses. Or you can use the sample service, just for test.

  • The example provides english model and chinese model, and Full-text search . If you want use Another language , you can train your language generation service.

  • Data service contain text-generator and full-text-search. The new repo will be on soon.

  • You can build a API service like : Post Json:

{
	"context": "Yes, We ",
	"token": "Your_token",
	"article_type": "english",
	"max_length": 10,
	"number": 3,
	"is_index": true
}

Response Json:


{
	"ref": [{
		"content": "...",
		"title": "Nothing"
	}],
	"sentenses": [{
		"value": ", the people of the United States, stand together"
	}, {
		"value": " to say, this is the best deal we've"
	}, {
		"value": " can't say anything, but it's not our"
	}]
}