Ontology-based protein-protein interaction explanation using large language models
Run llama2_finetuning.py Next, test the finetuned model with test.py Before running token_wise_metrics.py, get the responses generated with the testing (responses are the ones generated after ###Response:, sometimes model generates random outputs that does not follow the required output. Thus, we post-processed the generated outputs manually since we do not have much test sentences.) Fill the "Predicted" column in the excel file given as "finetuned_13b_chat.xlsx" ("13b" indicates the Llama-2-13b-chat model, change the file name according to your finetuned model)