Mixed cached dataset

Dataset that combines and handles multiple cached datasets, e.g. for multiple qubits. Here we also handle paddings.

source

Mixed_Cached_OpenClip_Dataset_config

 Mixed_Cached_OpenClip_Dataset_config (store_dict:dict, optimized:bool,
                                       dataset_to_gpu:bool,
                                       random_samples:int,
                                       num_of_qubits:int, min_gates:int,
                                       max_gates:int, gate_pool:list[str],
                                       pad_constant:int, collate_fn:str,
                                       bucket_batch_size:int,
                                       num_down_scales:int)

source

Mixed_Cached_OpenClip_Dataset

 Mixed_Cached_OpenClip_Dataset (device:torch.device=device(type='cpu'),
                                **parameters)

Dataset that uses multiple cached dataset and combines them with padding, either i) Bucket or ii) Max. Also provides a corresponding collate_fn for training.

Back to top