{
"dataset_param": {
"dataset_type": "lumina",
"basic_parameters": {
"data_config": {
"path": "/path/data.json"
}
},
"preprocess_parameters": {
"item_processor_type": "TokenizerItemProcessor",
"patch_size": 32,
"target_size": 256
},
"tokenizer_config": {
"hub_backend": "hf",
"autotokenizer_name": "luminaMGPT2Tokenizer",
"from_pretrained": "./weights/Lumina/Lumina-mGPT-2.0/"
}
},
"dataloader_param": {
"dataloader_mode": "sampler",
"sampler_type": "LuminaMetaLenDistSampler",
"shuffle": false,
"drop_last": true,
"pin_memory": true,
"group_frame": false,
"group_resolution": false,
"collate_param": {}
}
}