RFW0112: Tracking STT and TTS models and creating documentation
Summary
Tracking the model version and its hyperparameter and documenting of hyperparameter tuning and experiments
Documentation is for both Technical/Non-Technical
Key Concepts
hyperparameter:
A hyperparameter is a machine learning parameter that is chosen before a learning algorithm is trained. Hyperparameters are used to improve the learning of a model. They control how the model is training and determine the values of model parameters that a learning algorithm ends up learning
Context
We need a document to keep track of all the experiments and settings we've adjusted. This way, anyone can easily join in and help with the project. All the information regarding the model training and its results of the past should be clear such that the new developer would be able to smoothly carry the future work.
Outputs
documentation should include:
Standardised version naming (version of dataset)
model meta data and link for where its hosted(hugging face, github,...)
model working explanation and flowchart
Data content distribution (eg Buddhist literature )
Hyperparameter tuned
Evaluation Metric and History
Graphs
Inputs
The training data and benchmark data set with its distribution report.
The model that is being trained
Timeline
Specify the expected delivery date for the project.
References
Have the documentation for all AI models standardized and in same page to access.
RFW for MT documentation
kaldan007
commented
8 months ago
RFW0112: Tracking STT and TTS models and creating documentation
Summary
Tracking the model version and its hyperparameter and documenting of hyperparameter tuning and experiments Documentation is for both Technical/Non-Technical
Key Concepts
hyperparameter: A hyperparameter is a machine learning parameter that is chosen before a learning algorithm is trained. Hyperparameters are used to improve the learning of a model. They control how the model is training and determine the values of model parameters that a learning algorithm ends up learning
Context
We need a document to keep track of all the experiments and settings we've adjusted. This way, anyone can easily join in and help with the project. All the information regarding the model training and its results of the past should be clear such that the new developer would be able to smoothly carry the future work.
Outputs
documentation should include:
Inputs
The training data and benchmark data set with its distribution report. The model that is being trained
Timeline
Specify the expected delivery date for the project.
References
Have the documentation for all AI models standardized and in same page to access. RFW for MT documentation