Frequently Asked Questions

This article guides you through the Workspace, helps you understand platform features, and answers common issues you may encounter with storage, model uploads, data formats, devices, compilation, deployment, inference, and file management.

Use this FAQ as a quick reference to troubleshoot problems and navigate the system more efficiently.

Workspace limitations

Q: Is there a storage limit per user?
A: Yes. Each user has a 10 GB storage limit. Once you exceed the limit, you will no longer be able to create new projects or upload additional models.

Q: Where can I check my total storage usage?
A: The Workspace displays your current storage usage in the top-left corner.

Q: What does the “Free up storage space” message mean?
A: This message appears when you have reached your storage limit.
To continue working, delete unused items such as:

Old or unused projects
Previous builds

Supported model formats

Q: I am trying to upload a model file using the Upload Model option, but it is not uploading. What are the supported model formats?

A: The platform supports the following frameworks:

PyTorch

Framework	Case	Required Files	Remarks
PyTorch	Case 1: Single Model File	- `PT` or `*PTH` (complete model architecture + trained weights) - `input_shape.txt`	`input_shape.txt` file must only contain a valid input shape in one of the accepted formats below: Tuple: `(1, 3, 224, 224)` Array: `[1, 3, 224, 224]` CSV:`1, 3, 224, 224`
PyTorch	Case 2: Model Definition + Weights	- `MODEL.PY` (full model definition) - `PT` or `PTH` (weights) - `input_shape.txt`	- Must include `model.py` containing the full architecture definition - The class name in `MODEL.PY` must match the trained model - `input_shape.txt` uses the same accepted formats: Tuple: `(1, 3, 224, 224)` Array: `[1, 3, 224, 224]` CSV: `1, 3, 224, 224`

MODEL.PY Requirements:

  * Must contain the complete architecture class definition.
  * The architecture class name must match the one used during training.
  * No syntax errors—file must load properly.

Important

If you upload weights only (PT/PTH), you must also include model.py with the complete architecture class definition.
The architecture class name in model.py must match the name used during training.
Ensure that:
- MODEL.PY has no syntax errors.
- You can load the model without runtime errors in PyTorch.
- The input shape must match the model’s expected dimensions.

TensorFlow

Framework	Case	Required Files	Remarks
TensorFlow	Case 1: Model in H5 format	- `H5`	Complete model (architecture + weights).
TensorFlow	Case 2: Model in PB format	- `PB`	Complete model with architecture. Must represent a frozen graph.
TensorFlow	Case 3: Model in `SAVED_MODEL` directory format	- `SAVED_MODEL/` directory	Directory must contain the complete exported model (do not modify or delete internal files).

Important

Ensure the model is properly exported from TensorFlow.
Do not upload checkpoint-only or partial models.
The model must be loadable using standard TensorFlow methods:
H5 for full models,
PB for frozen graphs,
saved_model/ for full SavedModel exports (keep internal structure intact)

ONNX

Framework	Case	Required Files	Remarks
ONNX	Model in ONNX format	ONNX	Must contain the complete model graph and all trained parameters.

Warning

ONNX models must use opset version 13 or lower

Important

Export with all required operators included (no missing ops). Check the list of operators here.
Ensure ONNX version compatibility with your target runtime.
Verify the file is not corrupted or partially exported.

Q: What additional files are required based on task type?

A: Some tasks require extra metadata files. Refer to below requirements:

Additional Files (Based on Task Type)

Some tasks require extra metadata files so the system can interpret outputs correctly. Filenames are case sensitive and must match exactly.

Tasks requiring classes.txt:

Keyword Spotting
Genre Identification
Image Classification
Image Segmentation
Object Detection
Action Recognition
Visual Wake Word

classes.txt rules:

Rule	Description
One per line	Write one class label per line
No extras	Don’t include commas, brackets, extra spaces, or numbers (unless part of the label)
No empty lines	The file must not contain blank lines
Order matters	Class order must match the model’s output order
Count must match	The number of labels must equal the model’s output dimension

Example (classes.txt):

* cat
* dog
* car

Tasks requiring sampling_rate.txt:

Denoiser

sampling_rate.txt rules:

Rule	Description
First line only	Must contain one integer on the first line
No units	Do not include “Hz” or any text
No extra characters	No symbols or trailing spaces
Must match training	Value must match the training sampling rate

Example (sampling_rate.txt):

* 16000

Q: How does the system detect the framework of my BYOM model?
A: After uploading, the system automatically detects the framework and displays it in the Framework tab under both Create Project (BYOM) and Upload Model.

Q: Why is my PyTorch model not uploading?
A: Ensure that:

The file is in ZIP or TAR.GZ format
The file size is under 100 MB

Q: Why am I getting a “Missing Class” error during upload?
A: The selected task requires a class file included with the model package. Add the appropriate class file, then upload again.

Q: I've already created a PyTorch model and added the classes.txt file, but why is the upload still failing?

A: For PyTorch models, the input shape .txt file is mandatory. Even if the model file and classes.txt are correctly added, the upload will fail without a valid input shape file. Refer to below table for the requirements:

Input Shape File Requirements

Requirement	Details
File type	`.txt`
File name	Any valid name (`input.txt`, `shape.txt`, etc.)
Content	Must contain only a valid input shape and match the model's expected input size
Supported formats	Tuple: `(1, 3, 224, 224)` Array: `[1, 3, 224, 224]` CSV: `1, 3, 224, 224`

If the input shape file is missing or incorrectly formatted, the PyTorch model upload will fail.

Q: Where can I view my uploaded models?
A: Go to the My Models tab.

Supported input data formats

Q: Why can’t I upload certain file types on the Input Data page?
A: Only file types compatible with the selected task type and the compiled model are allowed. Files that do not meet the model’s requirements will be rejected.

Example: Vision models only accept image based file formats

Q: What input file formats are supported after compilation?
A: Supported formats vary by model and may include:

WAV
CSV
JPG
JPEG
PNG

Device selection

Q: Where can I view compatible devices for an ADI model?
A: Follow below steps to see the compatible devices for each model:

Open ADI's Model Zoo.
Select a model.
Go to the Model Details page.
Expand the Performance dropdown → view Compatible Platforms.

Q: Why are some devices not listed?
A: A device will appear only if it:

Is compatible with the selected model
Is online
Is sending a heartbeat signal (indicating active connection)

A heartbeat is a periodic signal indicating the device is online and ready.

Compilation

Q: Why is the Start Compilation button disabled after completing a compilation?
A: You cannot run another compilation with the exact same parameters. Modify at least one parameter to enable the button and start a new compilation.

Q: Why can’t I change parameters while a compilation is running?
A: Parameters are locked during an active compilation.
To make changes:

Abort the current compilation
Update the parameters
Restart the compilation

Q: How do I abort a compilation?
A: Click Abort Compilation during an active compilation.

Q: Can I restart an aborted compilation?
A: Yes. After aborting, simply click Start Compilation again.

Q: How can I find the reason for a failed compilation?
A: Open the Logs section on the Compilation page.

Pre- and post-processing scripts

Q: My Pre processing scripts are failing, what are the mandatory requirements for preprocessing scripts?

A: The pre-processing script must:

Include required imports:
argparse
Path
numpy
Accept command-line arguments:
--source
--output_file_path
--quant_type
Load raw input data.
Apply task-specific preprocessing.
Return a valid NumPy array.
Save output as .npy
Print output shape.

Note that the return value cannot be None.

Deployment

Q: Why can’t I start inference after uploading input files? Why are more steps required?
A: For BYOM models, you must complete:

Pre processing
Post processing

Inference can begin only after both steps complete successfully.

Q: Why can’t I proceed to Post-Processing after entering the Pre-Processing script?
A: The Pre-Processing script must run without errors. Check the Logs section for details.

Inference

Q: Where can I view my inference history?
A: Use the dropdown menu on the Inference Result page.

Q: Can I start another inference using the same compiled model?
A: Yes. Click Start New Inference on the Inference Result page.

Q: Can I download inference results?
A: Yes. Select Download PDF on the Inference Results page.

Q: How can I export inference results?
A: Use Export Inference on the Inference Results page.

File management

Q: Where can I view all builds under a project?
A: Open the project in the Project Workspace.

Q: How can I export a project?
A: In Project Workspace, use the Export Build option in the Project Status panel.

Q: Where can I view the projects I created?
A: Go to Workspace → Projects.

Q: How can I filter created models and projects?
A: Use the Filters panel on the left to filter by modality or framework.

Q: Where can I find all projects associated with a specific model?
A: Open the model in Workspace > Models tab and check the Associated Projects section.