LZK_USTC

ubuntu 上的4卡3090跑 vicuna13B的模型，模型转化后不能调用？ “out = model( input_ids=torch.as_tensor( [[token] if not sent_interrupt else output_ids], device=device ), use_cache=True, past_key_values=past_key_values if not sent_interrupt else None, )” 在输入问题提交给模型后，这行代码报错：typeError: 'model' object is not callable

Invalid argument: ValueError: could not broadcast input array from shape (512,500,3) int$ shape (512,512,3) Traceback (most recent call last): File "/home/user/lib_local/miniconda3/lib/python3.8/site-packages/tensorflow/python/ops/script_ops.py", line 249, in __call__ ret = func(*args) File "/home/user/lib_local/miniconda3/lib/python3.8/site-packages/tensorflow/python/autograph/impl/api.py",...

### Checklist - [X] 1. I have searched related issues but cannot get the expected help. - [ ] 2. The bug has not been fixed in the latest version....

### Motivation 我好像用internvl2_8B的代码训不了Intervl3_8B，但是internvl2_8B的代码训练internvl2.5_8B没有问题 ### Related resources _No response_ ### Additional context _No response_

[Bug] 2nd_finetune for internvl2.5 using lora ('resource_tracker: There appear to be %d ')

5

### Checklist - [X] 1. I have searched related issues but cannot get the expected help. - [X] 2. The bug has not been fixed in the latest version. -...

**Describe the bug** 我是从internvl的官方仓库里过来的，官网的训练代码依靠以下参数调节不同模块的冷冻与lora数： --use_llm_lora 4 \ #语言模型部分的lora --use_backbone_lora 4 \ #vit部分的lora --freeze_llm True \ #语言模型本体冻结（lora微调） --freeze_mlp False \ #桥接器不冻结 --freeze_backbone True \ #vit模型本体冻结（lora微调) 请问在swift仓库中这些参数仍然有效吗？如果无效的话设置什么参数能达到以上效果呢？