AI that can check PDF data content accuracy

[nenok2k5]

Hi po,

Is there an AI that can check the data accuracy of the text contents of a PDF file versus the image files?

 

[alist1986]

Yes, correct me I'm I'm wrong. Itong "cross-checking the data accuracy between text extracted from a PDF and that obtained from image files" ang ibig mong sabihin ha?
You do not have permission to view the full content of this post. Log in or register now.

Pagdating sa pdf extraction of text, ang gamit ng mga AI is the old OCR solutions such as used by Adobe Acrobat DC and Tesseract OCR (open source) application, na pwede mong gamitin separately for comparison. Hindi yan AI.

Pero yang latest AI models ng Google (as example) ay meron na nyang kakayahan. Kumpara mo na lang sa OpenAI models to see the difference. Sa ngayon halos lahat ng commercial AIs, meron na nyan. Iba-iba kasi ang construction ng mga transformer na yan, so there are differences in accuracy.

Ang standard features ng mga LLM and GPTs (transformers) vary kaya bubusisiin ninyo kung ano laman ng mga yan like:
1. tokenizers to convert text to tokens
2. embedding layers to convert tokens to vectors
3. transformer layers: for encoding and decoding
4. un-embedding layers: reverse noong #2
for text functions, TTS, STT, image, audio, video (etc.) api components sa service. Iikot sila sa capabilities ng AI for learning, reasoning, problem-solving, perception, and language understanding.

Yung ibang maliliit na LLMs na open source at mga old versions ng commercial AIs ay walang kakayahang bumasa ng pdf unless magdagdag ka ng modules to read a pdf, extract text from it, transcribe, summarize, etc and create a chatbot for that type of document. Ngayon, they are used as plugins to create AI agents for specific tasks. Mskikita mo mga yan sa ᑕᕼᗩTGᑭT playground that assists your model for whatever is was made to do - like pdf reading/summarizing.

 

[nenok2k5]

Thank you for the info!!