Comments (7)
Yes it requires large memories, the computational complexity of attention increase quadratically with the increase of the input length, you can reduce the length of track segmentation to reduce the memory cost.
from cmgan.
Thank you for your answer,I wonder if reducing the track length will affect the performance of the model?
from cmgan.
Yes, of course, it has some impact because attention relies on long-distance dependency if you reduce the track length to 40ms or 80ms it will have an extensive drop.
from cmgan.
Sorry to bother you again, I try to use the code training model to use batch_ size=4,cut_ len=16000 * 2, and the model trained for 80 and 120 epochs was tested, but the two results were found to be lower than the result in the paper.
from cmgan.
How much lower it is? If not much you can try to test with less epochs, as we train with only batch size of 3.
from cmgan.
the ssnr is only 10.4 in epoch=80, I want ask what method do you use to reduce the data to 16 khz.
from cmgan.
You can use the librosa downsampling which is similar to torch:
audio_down, sr = librosa.load(audio_path, sr=16000)
from cmgan.
Related Issues (20)
- RuntimeError HOT 3
- the change of gen_loss during training HOT 1
- RuntimeError
- RuntimeError HOT 2
- About the decreasing of loss HOT 1
- Can not reproduce the results HOT 12
- Training can get stuck HOT 6
- Inferior results trained from scratch HOT 7
- RuntimeeError HOT 1
- Can not reproduce the results HOT 3
- Training GPU requirements
- File "pesq/cypesq.pyx", line 1, in init cypesq ImportError: numpy.core.multiarray failed to import (auto-generated because you didn't call 'numpy.import_array()' after cimporting numpy; use '<void>numpy._import_array' to disable if you are certain you don't need it)
- File "/anaconda3/envs/cmg/lib/python3.9/site-packages/torch/nn/parallel/distributed.py", line 578, in __init__ dist._verify_model_across_ranks(self.process_group, parameters) RuntimeError: NCCL error in: ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:957, invalid usage, NCCL version 21.0.3 ncclInvalidUsage: This usually reflects invalid usage of NCCL library (such as too many async ops, too many collectives at once, mixing streams in a group, etc). HOT 2
- How do you resample to 16000? HOT 1
- 时域Loss计算疑惑
- the training speed confusion
- My server has a 3090, but reports that I don't have a gpu HOT 1
- Test set requirements when training
- epochs
- 模型训练的采样率以及显卡训练配置咨询
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from cmgan.