# Can I use LLMs to solve this challenge? **URL:** \ **Category:** SNOMED CT Entity Linking\ **Created:** [December 21, 2023, 11:49pm UTC](https://community.drivendata.org/t/can-i-use-llms-to-solve-this-challenge/9725 "2023-12-21T23:49:07Z")\ **Posts on this page:** 5\ **Page:** 1
**Author:** ![vaibhav2614](https://avatars.discourse-cdn.com/v4/letter/v/5e9695/32.png) [@vaibhav2614](https://community.drivendata.org/u/vaibhav2614)\ **Post date:** [December 21, 2023, 11:49pm UTC](https://community.drivendata.org/t/can-i-use-llms-to-solve-this-challenge/9725/1 "2023-12-21T23:49:07Z")
I can’t tell if I can solve this challenge using LLMs like GPT4. Prompts on top of GPT-4 might be a quick way to get good performance on this problem, but this code would typically be making calls to OpenAI, and I’m not sure if the rules allow for code that reaches out to an external LLM API. ---
**Author:** ![emily](https://avatars.discourse-cdn.com/v4/letter/e/8c91f0/32.png) [@emily](https://community.drivendata.org/u/emily)\ **Post date:** [December 22, 2023, 12:00am UTC](https://community.drivendata.org/t/can-i-use-llms-to-solve-this-challenge/9725/2 "2023-12-22T00:00:40Z")
Hi @vaibhav2614, solutions for this challenge must be open-source (per the [rules](https://www.drivendata.org/competitions/258/competition-snomed-ct/rules/#winner-license-type-open-source-license)), which prohibits the use of GPT4. In addition, the [code execution environment](https://www.drivendata.org/competitions/258/competition-snomed-ct/page/818/#runtime) does not have network access. Open-source LLMs that packaged up as part of the submission assets are allowed. ---
**Author:** ![raviteja\_b](https://avatars.discourse-cdn.com/v4/letter/r/e9c0ed/32.png) [@raviteja\_b](https://community.drivendata.org/u/raviteja_b)\ **Post date:** [January 8, 2024, 4:02am UTC](https://community.drivendata.org/t/can-i-use-llms-to-solve-this-challenge/9725/3 "2024-01-08T04:02:05Z")
> Open-source LLMs that packaged up as part of the submission assets are allowed. Let’s use the popular Llama 2 as an example. The code to train Llama 2 is not public, but the weights are. > **[Meta’s LLaMa 2 license is not Open Source - Voices of Open Source](https://blog.opensource.org/metas-llama-2-license-is-not-open-source/)** > > Meta is lowering barriers for access to powerful AI systems, but unfortunately, Meta has created the misunderstanding that LLaMa 2 is “open source” - it is not. Many popular open source LLMs are fine tunes based off Llama 2 (all legal to bundle, but may or may not be considered open source, as per different definitions of open source). Unquantized Llama 2 70b is ~ 130 GB. There are no size constraints to submissions? I may not even submit, but I am just listing potential issues of interpretation. ---
**Author:** ![mmiron](https://avatars.discourse-cdn.com/v4/letter/m/5daacb/32.png) [@mmiron](https://community.drivendata.org/u/mmiron)\ **Post date:** [January 8, 2024, 4:40pm UTC](https://community.drivendata.org/t/can-i-use-llms-to-solve-this-challenge/9725/4 "2024-01-08T16:40:06Z")
I don’t think an off-the-shelf solution will score well enough to make it an issue, personally, but I wouldn’t stake my life on it. While you could technically try using a 7b model, it likely wouldn’t even complete within the hour time limit (the runtime GPU has 80GB of RAM and an hour time limit). ---
**Author:** ![emily](https://avatars.discourse-cdn.com/v4/letter/e/8c91f0/32.png) [@emily](https://community.drivendata.org/u/emily)\ **Post date:** [January 8, 2024, 5:17pm UTC](https://community.drivendata.org/t/can-i-use-llms-to-solve-this-challenge/9725/5 "2024-01-08T17:17:31Z")
Hi @raviteja_b – if you need more clarity on what constitutes “open source,” you can check out the rules. As stated in the [rules](https://www.drivendata.org/competitions/258/competition-snomed-ct/rules/#data-use-and-code-sharing), prize-eligible submissions may **not** use: - software that is not licensed under an Open Source Initiative approved license; or - an open source license that prohibits commercial use Common OSI licenses are MIT and Apache 2.0. You can find a full list [here](https://opensource.org/licenses/).