ZipPy: Fast method to classify text as AI or human-generated
 
 
 
 
Go to file
Jacob Torrey 73997227bc Added code to make Nim compile to CLI and web
Signed-off-by: Jacob Torrey <jacob@thinkst.com>
2023-06-09 03:46:28 -06:00
.github/workflows Adding separate tests for nim code 2023-06-09 03:46:27 -06:00
nlzmadetect Added code to make Nim compile to CLI and web 2023-06-09 03:46:28 -06:00
samples Adding samples generated by chatGPT written in the voice of a highschooler 2023-06-09 03:46:27 -06:00
.gitignore Initial commit 2023-06-09 03:44:42 -06:00
.gitmodules Added code to make Nim compile to CLI and web 2023-06-09 03:46:28 -06:00
LICENSE Initial commit 2023-06-09 03:44:42 -06:00
README.md Update README.md 2023-06-09 03:46:27 -06:00
ai-generated.txt Adding samples generated by chatGPT written in the voice of a highschooler 2023-06-09 03:46:27 -06:00
lzma_detect.py Fix bug in finalize, and add one more generated abstract to ai-generated from samples 2023-06-09 03:46:27 -06:00
openai_detect.py Added OpenAI connector and pytest harness 2023-06-09 03:46:28 -06:00
roberta_detect.py Added code to use roberta-openai-detector to compare results against 2023-06-09 03:46:27 -06:00
roberta_local.py Added code to use roberta-openai-detector to compare results against 2023-06-09 03:46:27 -06:00
test_lzma_detect.py Improve performance by farming out chunks to other processes with multiprocessing 2023-06-09 03:46:27 -06:00
test_openai_detect.py Added OpenAI connector and pytest harness 2023-06-09 03:46:28 -06:00
test_roberta_detect.py Added code to use roberta-openai-detector to compare results against 2023-06-09 03:46:27 -06:00

README.md

ai-detect: Fast methods to classify text as AI or human-generated

This is a research repo for fast AI detection methods as we experiment with different techniques. While there are a number of existing LLM detection systems, they all use a large model trained on either an LLM or its training data to calculate the probability of each word given the preceeding, then calculating a score where the more high-probability tokens are more likely to be AI-originated. Techniques and tools in this repo are looking for faster approximation to be embeddable and more scalable.

LZMA compression detector (lzma_detect.py and nlzmadetect)

Python classifiation accuracy testing Nim classification accuracy testing

This is the first attempt, using the LZMA compression ratios as a way to indirectly measure the perplexity of a text. Compression ratios have been used in the past to detect anomalies in network data for intrusion detection, so if perplexity is roughly a measure of anomalous tokens, it may be possible to use compression to detect low-perplexity text. LZMA creates a dictionary of seen tokens, and then uses though in place of future tokens. The dictionary size, token length, etc. are all dynamic (though influenced by the 'preset' of 0-9--with 0 being the fastest but worse compression than 9). The basic idea is to 'seed' an LZMA compression stream with a corpus of AI-generated text (ai-generated.txt) and then measure the compression ratio of just the seed data with that of the sample appended. Samples that follow more closely in word choice, structure, etc. will acheive a higher compression ratio due to the prevalence of similar tokens in the dictionary, novel words, structures, etc. will appear anomalous to the seeded dictionary, resulting in a worse compression ratio.