Natural Language Processing Course

Automatic Generation of Slovenian Traffic News for RTV Slovenija

Running inference on HPC

1. Change to the project directory:

   cd /d/hpc/projects/onj_fri/groupid
   

2. Prepare input data by modifying or adding entries in the ./src/inputs.json file

3. Submit the inference job:

   sbatch ./jobs/inference.sh
   

4. Outputs will be saved (relative to the directoy you ran from) in the file: ./logs/inference-<job_id>.out

Finetuning on HPC

1. Change to the project directory:

   cd /d/hpc/projects/onj_fri/groupid
   

2. Submit the finetuning job:

   sbatch ./jobs/finetune.sh
   

3. (LORA) Adapter will be saved in the folder: ./models/