An AI-powered desktop application for generating educational fill-in-the-blank sentences.
LexiGen is an open-source fill-in-the-blank problems generator using local AI with Zero Cost.
Enter your words:
- Press "Generate" or use "Ctrl/Cmd + Enter" to clean the generated sentences and replace them with new ones
- Press "Append" or use "Enter" to append new sentences under the generated ones
Press Show/Hide, Show/Hide All or pressing "Ctrl/Cmd + /" to show/hide blanks in sentences.
Not satisfy with the sentences? Press "↻" to regenerate.
Press "⋮" next to the sentences to browse the menu:
- Move Up/Down: Move the current sentence upward or downward to order sentences.
- Analyze: Pop up a new window to analyze the function of the given word in the sentence
- Edit: Edit the sentence by yourself
- Designate: Designate a tense/mood to regenerate the sentence, supporting:
- [Present, Past, Future, Past Future] x [Simple, Continuous, Perfect, Perfect Continuous]
- Subjunctive Mood [Preeent, Past]
- Conditional [Zero, First, Second, Third]
- Imperative Mood
Want to share the sentence set? Press "Menu -> Export to Word" or "Ctrl/Cmd + E" to export all your sentences to a docx. You can choose to generate analysis.
Answer Key will be automatically appended to the last pages of the document.
Want your sentences generated under a certain context? Press "Context" or "Ctrl/Cmd + T" to add a context to all New Sentences. When you close the context window by the keyboard shortcut, the context will be automatically saved.
Press "Menu -> Save/Load" or "Ctrl/Cmd + S"/"Ctrl/Cmd + O" to save/load all current data, including words, sentences, analyses, and context to/from an external YAML file.
You can share YAML files with your friends/students/teachers!
If you do NOT want to download Ollama and use terminal to download models, go to "Alternative: Using Local GGUF Models".
-
Install Ollama
-
Launch the Ollama app
-
Open the terminal (Windows: press
Windows + R, then typecmd) and install a model (suggestingqwen2.5:3borqwen2.5:0.5b(smaller)):ollama pull qwen2.5:3b
LexiGen now supports using local GGUF models directly, without requiring Ollama:
- Download a GGUF model from Hugging Face and place it in the
LexiGenAssets/modelsdirectory - Launch LexiGen and:
- Set the API URL dropdown to "GGUF Models" (one of the default options)
- Select your downloaded model from the dropdown
"LexiGen_Version_GGUF_System.zip" packages in Releases already contains qwen2.5-1.5b-instruct-q4_k_m.gguf, under Apache 2.0 License.
- Download LexiGen from the Releases
- Run the installer for your platform
- Launch LexiGen
You can customize your prompts! Check the guide Customize Prompts.
LexiGen uses a dedicated LexiGenAssets folder for all external files:
LexiGenAssets/- The main assets directorytranslations/- Translation files (e.g., zh_CN.json)prompt.txt- Custom prompt template filesettings.yaml- Application settings
This structure separates application code from user data, making it easier to manage and backup your settings.
| Issue | Solution |
|---|---|
| No sentence generation | Check if Ollama server is running |
| Connection errors | Use 'Check Server' button |
| Slow performance | Verify system resources or use a smaller model qwen2.5:0.5b |
| Model errors | Confirm model installation |
"Lexi" comes from "lexicon".
I considered the name "FibGen" (Fill In the Blank). However, because "fib" means lie, this name was discarded.
Contributions are welcome! Here's how you can help under GNU General Public License v3.0:
- Report bugs in Issues
- Suggest features in Issues
- Post your POV to this project in Discussions
- Submit pull requests
If you like this project, please consider giving me a star ⭐️