NuExtract: A LLM for Structured Extraction

(huggingface.co)

16 points | by dmezzetti 302 days ago

3 comments

  • lmeyerov 302 days ago
    I've been mixed on this, it seems great for throwaway projects but a dead end for ones that survive:

    * Out-of-the-box fine-tuned model for extraction makes for an easier start, so happy to see this out there

    * ... But presumably one of the next steps in a project using it is doing further fine-tuning on their data after some use, but this is a closed model with closed data, so preserving the original fine tuning quality is unclear without that

    • etiennebcp 296 days ago
      Hi there, CEO of NuMind here, absolutely agree, these models are meant to be fine-tuned to be really useful. Right now you can see how the input -> output are formatted in the model card on HuggingFace. From that you can create a set of input -> output and fine-tune NuExtract. We are working on a platform to make this fine-tuning easier.
    • dmezzetti 302 days ago
      Fair points. GLiNER is another model to consider: https://huggingface.co/urchade/gliner_medium-v2
      • jerrygenser 302 days ago
        Gliner is cc-by-nc-4. Seems like it's not licensed for commercial use?
        • dmezzetti 302 days ago
          Sorry, I referenced the older model. The v2 models are Apache 2.0.
          • lmeyerov 299 days ago
            Neat, afaict, if someone fine-tunes for their own data, the gliner stack is OSS so they may be able to just fork & rerun/continue
    • jerrygenser 302 days ago
      Where are you getting that it's a closed model? I'm seeing mit license on huggingface.
      • lmeyerov 302 days ago
        Where is the fine-tuning source code + data and with what license? Afaict that's the proprietary part which is why it isn't in their github and hf doesn't link to it.

        Hence my comment that I don't see how to do further fine-tuning that doesn't just wipe out what this does. They did a free binary release of fine-tuned weights afaict, which is problematic for building a production data pipeline that you re-fine-tune over time.

        • jerrygenser 301 days ago
          I see. Yeah, I'd be interested in fine tuning as well on my own extractions. The tiny model even says it's meant to be fine tuned. I guess that's what they are commercializing :/
  • ranger_danger 302 days ago
    ELI5? I cannot tell what this is, what I'm supposed to do with it, or why it's good or newsworthy.
    • dmezzetti 302 days ago
      It's for NER (Named Entity Extraction). In the past, models had to be trained to know how to find entities such as names, addresses, dates. This model has the capability to do that zero-shot.

      In other words, you can pass it text then say extract the names, addresses and dates.

  • ganeshkrishnan 302 days ago
    are there any examples on how this works?