Comments (1)
Hi @chododom, TFTModel is a transformer model so it is more complex and more inefficient compared to the other models.
From my checks, increasing the hidden size from 16 to 32 (factor 2), increased the number of trainable params by a factor 4.
So increasing the hidden size from 8 to 10 (factor 1.25) leading to an increase in training time by a factor 11 does indeed sound strange (not saying yet that it is a bug though).
We would have to perform an in-depth analysis and profile the model to see whether this is normal. Currently, we don't have much capacity on our side for this as we're working on higher-prio tasks. So any help from the community would also be greatly appreciated :)
Also, we have some additional recommendations for model performance in our user guide for torch models.
from darts.
Related Issues (20)
- TSMixer ConditionalMixer Skip Connections HOT 1
- [BUG] PLForecastingModule._calculate_metrics should call log_dict with batch_size HOT 2
- [BUG] TorchMetrics loop implementation isn't compatible with stateful metrics HOT 1
- [BUG] TCNModel changes the shape of data erroneously before loss evaluation HOT 3
- [BUG] TFTModel returns StopIteration on decoder_vsn() call in forward() HOT 5
- How to make TorchForecastingModel store the best model checkpoint? HOT 1
- New model - TimeMixer
- [BUG] Static covariates not added to val_series for RegressionModel HOT 1
- Embedding extraction for clustering HOT 1
- [BUG] Scaler took a long time to start fit
- [BUG] Distributed prediction crash HOT 2
- Loading historical forecasts of future covariates for training - forecast changes each time step
- [Question] How to plot prediction in tensorboard for torch/lightning models?
- Enhance integration of Global and Local models. HOT 2
- [QUESTION] from darts.explainability import TFTExplainer ImportError: cannot import name 'TFTExplainer' from 'darts.explainability' HOT 1
- Is there a Way to Represent Static Covariates and Also Different Items? HOT 1
- Question about past covariates, input_chunk_length, output_chunk_length and using trained model in real-life problem. HOT 2
- [FEATURE REQUEST] Add temporal_hidden_past and temporal_hidden_future hyperparams to TiDEModel
- [BUG]Inconsistent Prediction Behavior Using GPU vs. CPU in Darts Framework HOT 1
- [Question] 'numpy.linalg.LinAlgError' with VARIMA model HOT 5
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 darts.