SAME DOCUMENT · TWO ENGINES
See where the results differ.
We ran the selected GitHub project on a published research abstract and checked the same text with the current website.
The sample
395 characters · 3 sentencesThe dominant sequence transduction models are based on complex recurrent or convolutional neural networks that include an encoder and a decoder. The best performing models also connect the encoder and decoder through an attention mechanism. We propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely.
Read the original: Attention Is All You NeedRecorded web trial
16/09/2026, 17:26:11 UTCThe copied abstract was not flagged. The original source was found, but each sentence scored below the project's 50% threshold.
Sentence 137%
The dominant sequence transduction models are based on complex recurrent or convolutional neural networks that include an encoder and a decoder
doi.org37%arxiv.org27%doi.org27%doi.org23%Sentence 235%
The best performing models also connect the encoder and decoder through an attention mechanism
doi.org35%arxiv.org27%Sentence 327%
We propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely
arxiv.org27%doi.org16%The submitted passage was located in retrieved source text. Overlapping matches are counted once.
These are different measurements: the GitHub project's headline counts sentences above a threshold; the current website counts verified text overlap. This single trial demonstrates behavior, not general detection accuracy.
TRY THE SCORING
Give both engines the same source.
Change the text or choose an example. This test runs the GitHub project's original scoring functions and the current matcher on the source you provide. It does not search the web.
Solar panels convert sunlight into electrical energy using semiconductor materials.
What this GitHub revision actually includes
Checked against the downloaded source code. Its README lists features that are absent from the current application.
| Capability | GitHub project | Current website |
|---|---|---|
| Text input | Yes | Yes |
| DOCX / PDF upload | Not present in this revision | Available |
| AI writing detection | Not implemented | Available; not part of this trial |
| PDF report download | Not present in this revision | Available |
| Document coverage | First 20 qualifying sentences | Up to 100,000 characters |
| Source text limit | First 5,000 characters per page | Up to 150,000 characters per source |
| Match location | Whole-sentence score | Exact matching text ranges |