训练这个模型在其他数据集 about mctrans HOT 8 CLOSED

------------------&nbsp;原始邮件&nbsp;------------------ 发件人: ***@***.***&gt;; 发送时间: 2022年3月30日(星期三) 晚上8:33 收件人: ***@***.***&gt;; 抄送: ***@***.***&gt;; 主题: [JiYuanFeng/MCTrans] 训练这个模型在其他数据集 (Issue #17) 你好，我将这个模型用于别的数据集（19类），并更改了配置文件里的类别个数，但似乎并不能直接运行起来，出现了下面的错误： Traceback (most recent call last): File "/mnt/csip-102/CaiJiaXin/MCTrans-master/train.py", line 131, in <module&gt; main() File "/mnt/csip-102/CaiJiaXin/MCTrans-master/train.py", line 127, in main SegPipline(cfg, distributed, not args.no_validate).run() File "/mnt/csip-102/CaiJiaXin/MCTrans-master/mctrans/pipline/segpipline.py", line 99, in run self.runner.run(self.data_loaders, self.cfg.workflow, self.cfg.max_epochs) File "/opt/conda/lib/python3.7/site-packages/mmcv/runner/epoch_based_runner.py", line 127, in run epoch_runner(data_loaders[i], **kwargs) File "/opt/conda/lib/python3.7/site-packages/mmcv/runner/epoch_based_runner.py", line 50, in train self.run_iter(data_batch, train_mode=True, **kwargs) File "/opt/conda/lib/python3.7/site-packages/mmcv/runner/epoch_based_runner.py", line 30, in run_iter **kwargs) File "/opt/conda/lib/python3.7/site-packages/mmcv/parallel/data_parallel.py", line 75, in train_step return self.module.train_step(*inputs[0], **kwargs[0]) File "/mnt/csip-102/CaiJiaXin/MCTrans-master/mctrans/models/segmentors/base.py", line 152, in train_step losses = self(**data_batch) File "/opt/conda/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1051, in _call_impl return forward_call(*input, **kwargs) File "/opt/conda/lib/python3.7/site-packages/mmcv/runner/fp16_utils.py", line 109, in new_func return old_func(*args, **kwargs) File "/mnt/csip-102/CaiJiaXin/MCTrans-master/mctrans/models/segmentors/base.py", line 122, in forward return self.forward_train(img, img_metas, **kwargs) File "/mnt/csip-102/CaiJiaXin/MCTrans-master/mctrans/models/segmentors/encoder_decoder.py", line 92, in forward_train loss_aux = self._auxiliary_head_forward(x, seg_label, return_loss=True) File "/mnt/csip-102/CaiJiaXin/MCTrans-master/mctrans/models/segmentors/encoder_decoder.py", line 81, in _auxiliary_head_forward return self.aux_head.forward_train(x, seg_label) File "/mnt/csip-102/CaiJiaXin/MCTrans-master/mctrans/models/heads/mctrans_aux_head.py", line 49, in forward_train outputs = self.ca(inputs_flatten) File "/opt/conda/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1051, in _call_impl return forward_call(*input, **kwargs) File "/mnt/csip-102/CaiJiaXin/MCTrans-master/mctrans/models/trans/transformer.py", line 145, in forward query = layer(query, src) File "/opt/conda/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1051, in _call_impl return forward_call(*input, **kwargs) File "/mnt/csip-102/CaiJiaXin/MCTrans-master/mctrans/models/trans/transformer.py", line 122, in forward tgt2 = self.cross_attn(tgt, src, src)[0] File "/opt/conda/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1051, in _call_impl return forward_call(*input, **kwargs) File "/opt/conda/lib/python3.7/site-packages/torch/nn/modules/activation.py", line 1038, in forward attn_mask=attn_mask) File "/opt/conda/lib/python3.7/site-packages/torch/nn/functional.py", line 5025, in multi_head_attention_forward k = k.contiguous().view(-1, bsz * num_heads, head_dim).transpose(0, 1) RuntimeError: shape '[-1, 152, 16]' is invalid for input of size 2752512 我看到关闭的问题中也有类似问题，但我按照他的修改方式也并不能成功，期待你的解答 — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***&gt;

------------------&nbsp;原始邮件&nbsp;------------------ 发件人: ***@***.***&gt;; 发送时间: 2022年3月30日(星期三) 晚上8:38 收件人: ***@***.***&gt;; 抄送: ***@***.***&gt;; ***@***.***&gt;; 主题: Re: [JiYuanFeng/MCTrans] 训练这个模型在其他数据集 (Issue #17) 我使用的配置文件是mctrans_vgg32_d5.py model = dict( type='EncoderDecoder', pretrained=None, encoder=dict( type="VGG", in_channel=3, init_channels=32, num_blocks=2), center=dict( type="MCTrans", d_model=128, nhead=8, d_ffn=512, dropout=0.1, act="relu", n_levels=3, n_points=4, n_sa_layers=6), decoder=dict( type="UNetDecoder", in_channels=[32, 64, 128, 128, 128], ), seg_head=dict( type="BasicSegHead", in_channels=32, num_classes=19, # 6 post_trans=[dict(type="Activations", softmax=True), dict(type="AsDiscrete", argmax=True)], losses=[dict(type="DiceLoss", softmax=True, to_onehot_y=True), dict(type="CrossEntropyLoss"), dict(type="FLoss")]), aux_head=dict( type="MCTransAuxHead", d_model=128, d_ffn=512, act="relu", num_classes=19, # 6 in_channles=[32, 64, 128, 128, 128], losses=[dict(type="MCTransAuxLoss", sigmoid=True, loss_weight=0.1)]), ) — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: ***@***.***&gt;