-
When i try to do transfer learning with latest master and asr_transducer, i am getting these warnings.
I suspect no weights are transferred.
I am doing:
--pretrained_model forkmodel.pth \
…
-
@danliu2 按照您的建议我重新训练了模型，在使用simuleval的时候遇到了翻译鬼畜的问题，使用的agent是text_fullytransducer_agent，我初步定位到翻译重复的地方是transducer_searcher文件里面infer函数内，在第一次调用self.searcher.search后产生的。
我打印了调用前后prev_tokens的tensor如下：
> …
-
hello. I have train a "LSTM-Transducer-BPE" model some days ago. and to save the model, I have used checkpoint. for this train, I have used the default config.
after that, when i want to load this mo…
-
Hi again,
I have tried to train your model without ASR pre-training and KD but it did not perform very well, as expected.
So, I pre-train the encoder using the script you provided and I perform the …
-
**Background information**: I use recipes in wenetspeech for training. There's no problem with oom in scan_pessimistic_batches_for_oom, but it's always oom in the later training. it will not work …
-
Hi I am installing k2 from source, but encountered the following error:
my pytorch installation commands
```
nvcc -V # 10.1
conda install pytorch==1.7.1 torchvision==0.8.2 torchaudio==0.7.2 cud…
-
I use pytorch profiler (https://pytorch.org/docs/stable/profiler.html) to compare relative contribution of different modules of the Conformer model to the taining time.
Code and more details can b…
-
After using the fast_rnnt loss in my environment， the trainning loss always failed into nan or inf.
The configuration fo my ConformerTransducer enviroment is as follows:
- v100-32g-4gpu * 2
- pla…
-
@csukuangfj
1. I tried to use beam-search with latest code then came up with bellow error. My Torch version is 1.11.0+cu113.
2. Then i tried to use CPU for decoding and the beam-search worked fi…
-
`python3 -m torch.utils.collect_env` outputs
```
Collecting environment information...
PyTorch version: 1.12.0+cu116
Is debug build: False
CUDA used to build PyTorch: 11.6
ROCM used to build PyT…