Skin-Cancer-Specialist-LLM

Fine tuning an LLM by the given dataset. The dataset represents the domain knowledge that the LLM should be an expert for.

The given notebooks are self-suffcient for fetching data (crawling), and fine-tuning an LLM (Llama, Mistral, Gemma, etc.) for training a domain knowledge.

Step 1:

You can use Fetching-Data-from-PubMed.ipynb for fetching the dataset by the given keywords. Such notebook helps find PubMed papers undet that topic for fetching.

Step 2:

After fetching the data, you need to put this into an instruction-input-output format in a json file to let it be trained via Fine-Tuning UnSloth on PubMed Articles.ipynb. One explanatory example can be PubMedAbstracts.json.

Step 3:

You can insert prompts for getting outputs from the questions you ask.