testxlog

testxlog

HomeArchives

How to use Obsidian to access the free AI writing assistant

241
AI Translation
This post is translated from Chinese into English through AI.View Original
AI-generated summary
The project is about using AI to assist in writing in Obsidian. There are two methods: using the MantouAI plugin with Alibaba Cloud's free model, or using the local-gpt plugin with local gemma models. The MantouAI method requires obtaining a Dashscope API Key and selecting functions like editing, translating, summarizing, or questioning. The local-gpt method involves downloading ollama software, installing gemma models, and setting up the local-gpt plugin. The MantouAI method is recommended for better Chinese language support, while the local-gpt method is suitable for privacy concerns but requires higher computer specifications.
  • Ideas
    • One is to use the MantouAI plugin to call the Alibaba Cloud limited-time free Tongyi large model, which has better support for Chinese in domestic models
    • The other is to use the local-gpt plugin to call the local gemma model repository, running models downloaded locally such as gemma:2b or qwen:4b
  • Methods
    • MantouAI Plugin Method
      • Install the MantouAI community plugin in Obsidian
      • Apply for a limited-time free Dashscope API Key from Alibaba Cloud
      • Fill in the obtained Key in the MantouAI plugin box
      • Provided that some text in the selected note is used as the corpus
      • Press ctrl+P to open the command panel and enter MantouAI
      • Choose functions like polishing, translation, summarization, questioning, etc., as needed
      • The result will be generated at the cursor position
        • Common functions can also be added to Obsidian shortcuts, such as setting the questioning function to ctrl+shift+?
    • Local-gpt Plugin Method
      • Download and install the ollama software from the official website on the computer system
      • Open the command line and enter ollama --version to confirm the correct installation version
      • Continue with ollama run gemma:2b to download and install the gemma:2b small model that supports 8GB of memory
        • Depending on the VRAM and memory size, multiple models can be downloaded and installed, such as:
        • Optional ollama run gemma:4b
        • Optional ollama run gemma:7b
        • Optional ollama run qwen:4b
        • Optional ollama run qwen:14b
        • Specific supported models can be queried from the ollama official website
      • Install the local-gpt plugin in Obsidian and configure it
        • Selected AI provider: Ollama
        • Set Ollama URL to http://localhost:11434
        • Choose the Default model based on the installed models, such as gemma:2b
        • Set a shortcut key for the local-gpt plugin, such as ctrl+alt+?, to differentiate from MantouAI's ctrl+shift+?
        • Provided that some text in the selected note is used as the corpus, press the ctrl+alt+? shortcut key
        • Select the desired command from the pop-up Summarize and confirm
          • Also, modify or add command names and prompts in the local-gpt plugin settings
  • Conclusion
    • Prioritize using the MantouAI plugin method for better Chinese support in domestic models
    • When privacy is a concern, the local-gpt plugin method can be used, but it requires higher computer specifications
Loading...
Ownership of this post data is guaranteed by blockchain and smart contracts to the creator alone.