AI proofreading with diff
Compare the source with an AI proofreading draft.
AI proofreading with diff
Cara menggunakan AI proofreading with diff
AI proofreading with diff sends your text to a small language model running in your browser, asks for a corrected version that keeps the meaning and language, and shows a word-level comparison of every change against the original. Download the corrected text as corrected.txt. The model downloads on first use, needs WebGPU, and accepts up to 8,000 characters.
- Paste the source into the input editor, or load the built-in example.
- Set processing.
- Run ai proofreading with diff, review the output, then use the available copy or download controls.
Fungsi alat ini
Downloads an inference model on first run. Chat and generation require WebGPU; smaller recognition models use browser WebAssembly. AI output can omit or invent details. Review it before use.
- Processing
- Local browser
Had dan pemprosesan
Chat sources: 8,000 characters; generation: up to 1,400 output tokens. First model downloads can be hundreds of MB. WebGPU chat and browser memory requirements vary by device. Model output is a draft.
Contoh input
A useful tool makes everyday work easier. Review the results before sharing.
Soalan lazim
How can I see what the AI changed in my text?
After the model returns its correction, the tool compares it with your original word by word and marks insertions and deletions. Review each change; the model is told to preserve meaning, but small models can still alter wording or facts.
Does the AI proofreader explain its corrections?
No. It is instructed to return only the corrected text, without explanations or headings, so the word diff is the record of what changed. Judge each edit yourself before using the corrected version.
Can I proofread text that is not in English?
The instruction asks the model to preserve the source language, so it attempts corrections in that language. Quality depends on how well the small local model knows it, which makes the word diff especially useful for checking edits in other languages.
Where is my input processed?
Local mode downloads public model/runtime assets and keeps input in this browser. Models may remain cached on this device.
What are the input limits?
Chat sources: 8,000 characters; generation: up to 1,400 output tokens. First model downloads can be hundreds of MB. WebGPU chat and browser memory requirements vary by device. Model output is a draft.