{
"dataset_param": {
"dataset_type": "deepseekvl2",
"basic_parameters": {
"data_path": "/home/ci_resource/data/internvlv1-2-sft/opensource/ai2d_train_12k_dsvl.jsonl",
"data_folder": "/home/ci_resource/data/internvlv1-2-sft/data/ai2d",
"repeat_time": 1
},
"processor_path": "/home/ci_resource/models/DeepSeelVL2/hf_model/deepseek-vl2",
"preprocess_parameters": null,
"group_by_length": false
},
"dataloader_param": {
"dataloader_mode": "sampler",
"shuffle": false,
"drop_last": true,
"pin_memory": true,
"sampler_type": "BaseRandomBatchSampler",
"collate_param": {
"model_name": "deepseekvl2",
"pad_id": 2
}
}
}