molywyp

Results 5 issues of molywyp

num_classes 为什么要+1

3 comment

Backbone部分没有载入是错误的。

2 comment

PS D:\segformer-pytorch> & d:/anaconda/python.exe d:/segformer-pytorch/train.py Load backbone weights Load weights model_data\segformer_b5_backbone_weights.pth. Successful Load Key: [] …… Successful Load Key Num: 0 Fail To Load Key: ['patch_embed1.proj.weight', 'patch_embed1.proj.bias', 'patch_embed1.norm.weight', 'patch_embed1.norm.bias', 'patch_embed2.proj.weight',...

系统尝试删除文件时遇到了权限问题

回溯（最近一次调用）： 文件 "d:\segformer-pytorch\train.py", 第 521 行, 在 中 fit_one_epoch(model_train, model, loss_history, eval_callback, optimizer, epoch, epoch_step, epoch_step_val, gen, gen_val, UnFreeze_Epoch, Cuda, \ 文件 "d:\segformer-pytorch\utils\utils_fit.py", 第 154 行, 在 fit_one_epoch 中 eval_callback.on_epoch_end(epoch...

max_new_tokens missing in generate() call in blip2_vicuna_instruct.py

Hi, thank you for your great work on MA-LMM! I noticed a potential issue in `MA-LMM/MA-LMM-main/lavis/models/blip2_models/blip2_vicuna_instruct.py`. In the current code, the `self.llm_model.generate()` method is called without specifying `max_new_tokens`: ```python outputs...

Missing import warnings in caption_builder.py

Hi, thanks for the great work! I noticed that the file `MA-LMM/MA-LMM-main/lavis/datasets/builders/caption_builder.py` is missing an import statement for the `warnings` module. This causes a `NameError` when calling `warnings.warn(...)`. To fix...