Before submitting an inference request for generative-question-answering, trainingStatus and inferenceStatus must be validated.
- Confirm that the training process has successfully concluded ->
trainingStatus:completed - Ensure that the inference model is in an operational state ->
inferenceStatus:running
The inferenceStatus can take three different values:
- Pending
- Running
- Null (Training is not done yet)
For models whose training is initiated via the API, the "Check Training & Inference Status" request serves as the trigger for publishing of the trained model. This publication process subsequently enables inference on the trained model.
Token Endpoint:
URL: {{baseUrl}}/sestekai-api/api/external/generative-question-answering-trainings/:modelId/progress
HTTP Method: GET
Content-Type: none
Here's an example of a request:
curl --location '{{baseUrl}}/sestekai-api/api/external/generative-question-answering-trainings/:modelId/progress' \
--header 'Authorization: Bearer <token>'
Expected Response:
{
"percent": 100
"trainingStatus": "Completed"
"inferenceStatus": "Running"
}
Potential training statuses that can be obtained for the training via external api:
| Training Status | Description |
|---|---|
| Initialized | Initial creation of the experiment. |
| PendingTraining | Train pod is being created in Kubernetes. Upon successful initiation, it transitions to DatasetUploading. |
| DatasetUploading | Dataset files are starting to be uploaded. |
| DataProcessing | Processing data once the files are successfully uploaded. |
| TrainingStarted | Confirms the initiation of the training process. |
| Training | Ongoing status of the training process. |
| Trained | Completion of the training process |
| ModelUploading | Initiation of uploading the trained model to Minio |
| Completed | Successful conclusion of all training stages. The associated Train pod is then terminated. |
| Failed | Occurrence of an error within the Train pod (AI module) |