预训练模型替换后不匹配救助

预训练模型替换成pretrained-11-7-21_75k大佬提供的出现不匹配。

Found 266 samples
+----------------+------------+---------------+------------------+
| Steps with r=2 | Batch Size | Learning Rate | Outputs/Step (r) |
+----------------+------------+---------------+------------------+
| 85k Steps | 12 | 5e-06 | 2 |
+----------------+------------+---------------+------------------+

E:\MockingBird\synthesizer\synthesizer_dataset.py:84: UserWarning: Creating a tensor from a list of numpy.ndarrays is extremely slow. Please consider converting the list to a single numpy.ndarray with numpy.array() before converting to a tensor. (Triggered internally at C:\cb\pytorch_1000000000000\work\torch\csrc\utils\tensor_new.cpp:248.)
embeds = torch.tensor(embeds)
E:\MockingBird\synthesizer\synthesizer_dataset.py:84: UserWarning: Creating a tensor from a list of numpy.ndarrays is extremely slow. Please consider converting the list to a single numpy.ndarray with numpy.array() before converting to a tensor. (Triggered internally at C:\cb\pytorch_1000000000000\work\torch\csrc\utils\tensor_new.cpp:248.)
embeds = torch.tensor(embeds)
Traceback (most recent call last):
File "synthesizer_train.py", line 37, in
train(**vars(args))
File "E:\MockingBird\synthesizer\train.py", line 208, in train
optimizer.step()
File "C:\Users\Administrator\anaconda3\envs\mock\lib\site-packages\torch\optim\optimizer.py", line 280, in wrapper
out = func(*args, **kwargs)
File "C:\Users\Administrator\anaconda3\envs\mock\lib\site-packages\torch\optim\optimizer.py", line 33, in _use_grad
ret = func(self, *args, **kwargs)
File "C:\Users\Administrator\anaconda3\envs\mock\lib\site-packages\torch\optim\adam.py", line 141, in step
adam(
File "C:\Users\Administrator\anaconda3\envs\mock\lib\site-packages\torch\optim\adam.py", line 281, in adam
func(params,
File "C:\Users\Administrator\anaconda3\envs\mock\lib\site-packages\torch\optim\adam.py", line 446, in _multi_tensor_adam
torch._foreach_add_(device_exp_avgs, device_grads, alpha=1 - beta1)
RuntimeError: The size of tensor a (1024) must match the size of tensor b (3) at non-singleton dimension 3

**Env & To Reproduce[复现与环境]**
描述你用的环境、代码版本、模型
Python 3.8.16 pretrained
**Screenshots[截图（如有）]**
If applicable, add screenshots to help

Reproduce the failure with the stated Python 3.8.16 environment and replacement pretrained model. Start at synthesizer/synthesizer_dataset.py:84 and synthesizer/train.py:208, then inspect the tensor shapes reaching optimizer.step(). Done means training no longer raises the reported tensor-size mismatch.

Written by the indexing model from the issue text.