Method __init__

data.py:470–492

(self, train_file, tokenizer, max_len=2048, sample=-1, test=False, seed=0, category="", K=4, dedup=False)

class SidSFTDataset_GPR(CSVBaseDataset):
    def __init__(self, train_file, tokenizer, max_len=2048, sample=-1, test=False, seed=0, category="", K=4, dedup=False):
        super().__init__(train_file, sample, seed, max_len, category, dedup, tokenizer, test)

        # Try to load features from standard location
        try:
            with open(f'data/{category}/{category}.user.json', 'r') as f:
                self.user_features = json.load(f)
        except FileNotFoundError:
            try:
                dataset_dir = os.path.dirname(train_file)
                # Assuming structure data/Amazon/train/Sports... -> data/Sports/Sports.user.json
                with open(f'data/{category}/{category}.user.json', 'r') as f:
                    self.user_features = json.load(f)
            except:
                self.user_features = {}
            
        try:
            with open(f'data/{category}/{category}.item.json', 'r') as f:
                self.item_features = json.load(f)
        except FileNotFoundError:
            self.item_features = {}
            
        self.get_inputs()  

    def get_history(self, row):
        row['history_item_sid'] = eval(row['history_item_sid'])