class InstructionDataset(Dataset):
    def __init__(self, data, tokenizer):
        self.data = data
        self.encoded_texts = []

        for entry in data:
            instruction_plus_input = format_input(entry)
            response_text = f"\n\n### Response: \n{entry['output']}"
            full_text = instruction_plus_input + response_text
            self.encoded_texts.append(tokenizer.encode(full_text))

    def __getitem__(self, index):
        return self.encoded_texts[index]

    def __len__(self):
        return len(self.data)

        __
        __