FineTune with new label.json

[masoud parpanchi]

I'm trying to finetune the librispeech model to train a Persian STT.
this is the command :
CUDA_VISIBLE_DEVICES=1 python train.py data.train_path=/mnt/ssd/masoud_parpanchi/cv-corpus-6.1-2020-12-11/fa/clips/train_data.json data.val_path=/mnt/ssd/masoud_parpanchi/cv-corpus-6.1-2020-12-11/fa/clips/dev_data.json trainer.gpus=1 data.batch_size=4 trainer.resume_from_checkpoint=/mnt/ssd/masoud_parpanchi/deepspeech2/checkpoint_libriSpeech/librispeech_pretrained_v3.ckpt

But I get the following error

Error executing job with overrides: ['data.train_path=/mnt/ssd/masoud_parpanchi/cv-corpus-6.1-2020-12-11/fa/clips/train_data.json', 'data.val_path=/mnt/ssd/masoud_parpanchi/cv-corpus-6.1-2020-12-11/fa/clips/dev_data.json', 'trainer.gpus=1', 'data.batch_size=4', 'trainer.resume_from_checkpoint=/mnt/ssd/masoud_parpanchi/deepspeech2/checkpoint_libriSpeech/librispeech_pretrained_v3.ckpt']
Traceback (most recent call last):
File "/mnt/ssd/masoud_parpanchi/deepspeech2/deepspeech.pytorch/train.py", line 21, in hydra_main
train(cfg=cfg)
File "/mnt/ssd/masoud_parpanchi/deepspeech2/deepspeech.pytorch/deepspeech_pytorch/training.py", line 53, in train
trainer.fit(model, data_loader)
File "/home/shenasa/anaconda3/envs/deepspeech2/lib/python3.9/site-packages/pytorch_lightning/trainer/trainer.py", line 499, in fit
self.dispatch()
File "/home/shenasa/anaconda3/envs/deepspeech2/lib/python3.9/site-packages/pytorch_lightning/trainer/trainer.py", line 546, in dispatch
self.accelerator.start_training(self)
File "/home/shenasa/anaconda3/envs/deepspeech2/lib/python3.9/site-packages/pytorch_lightning/accelerators/accelerator.py", line 73, in start_training
self.training_type_plugin.start_training(trainer)
File "/home/shenasa/anaconda3/envs/deepspeech2/lib/python3.9/site-packages/pytorch_lightning/plugins/training_type/training_type_plugin.py", line 114, in start_training
self._results = trainer.run_train()
File "/home/shenasa/anaconda3/envs/deepspeech2/lib/python3.9/site-packages/pytorch_lightning/trainer/trainer.py", line 602, in run_train
self._pre_training_routine()
File "/home/shenasa/anaconda3/envs/deepspeech2/lib/python3.9/site-packages/pytorch_lightning/trainer/trainer.py", line 593, in _pre_training_routine
self.checkpoint_connector.restore_weights()
File "/home/shenasa/anaconda3/envs/deepspeech2/lib/python3.9/site-packages/pytorch_lightning/trainer/connectors/checkpoint_connector.py", line 73, in restore_weights
self.restore(self.trainer.resume_from_checkpoint, on_gpu=self.trainer._device_type == DeviceType.GPU)
File "/home/shenasa/anaconda3/envs/deepspeech2/lib/python3.9/site-packages/pytorch_lightning/trainer/connectors/checkpoint_connector.py", line 100, in restore
self.restore_model_state(model, checkpoint)
File "/home/shenasa/anaconda3/envs/deepspeech2/lib/python3.9/site-packages/pytorch_lightning/trainer/connectors/checkpoint_connector.py", line 124, in restore_model_state
model.load_state_dict(checkpoint['state_dict'])
File "/home/shenasa/anaconda3/envs/deepspeech2/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1406, in load_state_dict
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for DeepSpeech:
size mismatch for fc.0.module.1.weight: copying a param with shape torch.Size([29, 1024]) from checkpoint, the shape in current model is torch.Size([39, 1024]).

I Saw there was a finetune flag before but now How to do this? The English model has 29 chars but I have 39.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.