CYL0089

Training time so long

18

hello, I found that the training time is so long(about 64 hours, 2.5 days). I have built the training code based on the official code of Tinyvit and your network...

OP "Resize" cannot convert when export tflite ? 转tflite 模型时Resize有问题

5

 ERROR: input_onnx_file_path: work_dirs/yolo_world_v2_s_vlpan_bn_2e-4_80e_8gpus_mask-refine_finetune_coco_ep80-492dc329.onnx ERROR: onnx_op_name: wa/baseModel/neck/upsample_layers.0/Resize ERROR: Read this and deal with it. https://github.com/PINTO0309/onnx2tf#parameter-replacement ERROR: Alternatively, if the input OP has a dynamic dimension, use the -b or -ois...

 NotImplementedError: this Node ScaledDotProductAttention-/baseModel/backbone/layers.2/layers.2.1/global_rep/global_rep.0/attn/ScaledDotProductAttention has no value_infer hi, I meet this error when I use an .onnx of MobileVIT to run the onnx_tool. And I don't kown how to...

 试了很多次都是这样，epoch是3，任务监控显示GPU利用率已经是0，请问这是为什么？ 训练log:  训练配置如下（--num_train_epochs 设置的 3）： ` "--model_name_or_path", "/mnt/Qwen2.5-VL-7B-Instruct", "--dataset_use", "T4LPD_No3", "--data_flatten", "True", "--tune_mm_vision", "True", "--tune_mm_mlp", "True", "--tune_mm_llm", "False", "--bf16", "--output_dir", "./output/T4LPD_No3_v5", "--num_train_epochs", "3", "--per_device_train_batch_size", "2", "--per_device_eval_batch_size", "0", "--gradient_accumulation_steps",...