In this projects, we have used some libaraies to help us pretrain the model. We first need to run
bash install.sh
to download required libaraies.
We need to first run the DNS
python3 idunno.py dns
Then, we can run the client, which would also turn on the coordinator, standby and worker
python3 idunno.py client
After run the project, we have following commands we can input after ">>>" in the terminal
To train the model
train model_name
For now, we can only use the model from Microsoft in hugging face
To upload inference date to simple distributed file system
upload input directory n
where n is the number of files of input directory we want
To infrerece,
inference model_name input_directory batch_size
To see the curry query rate of models and the number of queries processes so far for each models
C1
To see current processing time of a query of models
C2
To see the current set of VMs assigned to each model/job
C5