Skip to content

CFSAI build command

The cfsai build command compiles an AI model for an embedded target. It generates C or C++ source and headers, depending on the target, that allow the model to be used from within an embedded application.

Before you start

For full compatibility when using this feature, create a new workspace and select an AI-based template (for example, Zephyr Single Core TensorFlow AI Model for MAX32690). You can then run cfsai build to compile the defined models, or rebuild them with new parameters.

For details on supported processors, see supported processors and model formats.

Build options

  • Build with a .cfsconfig file. The configuration file can define one or more models along with the target SoC, core, and firmware platform.
  • Build a single model using required flags. Provide the model path with --model and the target with --target.

Note

When you use a .cfsconfig file with the --config flag, its settings take precedence and override any conflicting command-line options. The terminal output confirms this by displaying a message such as: Using .cfs/max78002-csbga.cfsconfig instead.

Build with a .cfsconfig file

Use a .cfsconfig file to build one or more models in a single command. The .cfsconfig file defines models, target SoC, core, and firmware platform.

cfsai build --config <path/to/project.cfsconfig>

Example

cfsai build -c .cfs/max78002-csbga.cfsconfig

Sample output

Created file: m4/src/adi_tflm/HelloWorldF32.cpp (OK)
Created file: m4/src/adi_tflm/HelloWorldF32.hpp (OK)

Generate a .cfsconfig file before you get started

To generate a .cfsconfig file, create a workspace using an AI-enabled template in CodeFusion Studio. The generated file appears in the .cfs/ folder of your workspace. Use CodeFusion Studio to configure the project and AI models. Look for the AIModels section in the file to see how models are defined.

Build using flags

Supply the model path with --model (or -m) and the target string with --target (or -t). The target format is <soc>?([<package>]).<core>?(.<accelerator>). Only one model is compiled per command using this method.

Component Description Example
soc System-on-Chip name MAX78002, MAX32690, ADSP-SC835
package (optional) Package variant of the SoC WLP, TQFN
core Processor core CM4, FX, CM33
accelerator (optional) Hardware AI accelerator CNN 
cfsai build --model <path/to/model_file> --target <soc>?([<package>]).<core>?(.<accelerator>)

C/C++ build compatibility

The cfsai build command generates C++ source and header files (.cpp, .hpp) for TensorFlow Lite Micro (TFLM) models. If your project is based on a non-AI template that uses a main.c source file, you may see compiler warnings when integrating the generated files. Use an AI-based template (for example, Zephyr Single Core TensorFlow AI Model), or replace main.c with main.cpp or a C++ wrapper.

Example

cfsai build --model model.tflite --target MAX78002.CM4.CNN
cfsai build --model model.tflite --target MAX78002.CM4
cfsai build --model model.tflite --target MAX32690[WLP].CM4
cfsai build --model model.tflite --target ADSP-SC835.FX

macOS zsh syntax

On macOS, zsh interprets square brackets as pattern tokens. Enclose the target in quotes to avoid a shell error: 

cfsai build --model model.tflite --target "MAX32690[WLP].CM4"

Tip

Run cfsai list-targets to see the complete list of supported combinations in your environment.

The --model flag accepts both local file paths and remote URLs. Remote files are cached for offline reuse and refreshed after one hour if the URL is available. Use cfsai clean-cache to clear the cache and force a re-download.

Additional flags

These flags let you control build output, select a firmware platform, or provide Izer-specific configuration.

Flag Description
--cwd Change working directory path. This is used as the base directory for relative file paths and project checks.
--extension, -e Provide backend-specific fields as key=value pairs.
--firmware-platform Select the firmware platform when a device supports multiple platforms. Typical values include msdk, xos, zephyr, and cces.
--izer-network-config Set the path or URL to the izer network configuration YAML file. Required when targeting the MAX78002 CNN accelerator.
--no-path-checks Disable checks related to project paths. Useful for CI pipelines or testing builds.
--only-core Generate output for the specified core only. The core name must match the CoreId field defined in the .cfsconfig file. This value is case-sensitive.
--output-path, -o Set the path to the directory where output files will be written. The default directory is the current working directory.

Common options

Run the build from a different working directory

Use a custom working directory as the base path for relative files. 

cfsai build --config .cfs/max78002-csbga.cfsconfig --cwd path/to/project-root

Write output to a custom folder

Place generated files in a directory other than the default.

cfsai build --config .cfs/max78002-csbga.cfsconfig --output-path ./custom_folder

Build output for only one core (must be defined in .cfsconfig)

Limit output to the specified core (must match a CoreId in .cfsconfig).

cfsai build --config .cfs/max78002-csbga.cfsconfig --only-core CM4

Disable path checks

Skip validation of project paths. Useful when building outside of a workspace.

cfsai build --config .cfs/max78002-csbga.cfsconfig --no-path-checks

Platform selection

Set a supported firmware platform

Choose the firmware platform when a device supports multiple platforms.

cfsai build --model path/to/model.tflite --target max32690.cm4 --firmware-platform msdk

CNN accelerator (Izer backend)

Set Izer network config

Provide a network configuration YAML file (path or URL) when targeting the MAX78002 CNN accelerator.

cfsai build \
    --target max78002.cm4.cnn \
    --model path/to/model.pth.tar \
    --izer-network-config path/to/config.yaml 
    # Alternatively, use --izer-network-config https://example.com/network.yaml

Note

For reference on how to train PyTorch models compatible with MAX78002, see the open-source ai8x-synthesis repository.

Provide backend-specific extensions

Pass custom flags defined by the backend to modify generation behavior.

cfsai build \
    --target MAX78002.CM4.CNN \
    --model path/to/model.pth.tar \
    --backend izer \
    -e Softmax=True

Note

The --backend izer flag must be specified if passing extensions.

Tip

Use cfsai list-backends, and cfsai list-extensions <backend> to discover supported backends and extension fields.

If using your own custom network configuration file, pass the model’s input shape using the --extension option.

cfsai build \
    --target MAX78002.CM4.CNN \
    --model path/to/model.pth.tar \
    --izer-network-config path/to/config.yaml \
    --backend izer \
    -e InputShape=1,32,32

Tip

The -e or --extension flag can be repeated to pass multiple key–value pairs. For example: -e InputShape=1,32,32 -e Softmax=True.

Global output options

Enable verbose logging

Print additional details while running a command. Helpful for debugging or understanding exactly what the tool is doing.

cfsai --verbose build --model model.tflite --target MAX32690.CM4

Sample output

🔍    Refreshing datamodel manager cache                                        
🔍    Found system package manager index                                                  
🔍    Found a system datamodel index 
🔍    Getting datamodel for MAX78002                         
🔍    Reading datamodel                                                          
🔍    Successfully validated configuration                                 
Created file "tensforlfow/src/adi_tflm/hello_world_f32.cpp"                        
Created file "tensforlfow/src/adi_tflm/hello_world_f32.hpp"

Emit JSON logs and results

Output machine-readable logs. Useful for CI pipelines or script parsing.
The output is in JSON Lines format (one JSON object per line), not a single JSON array.

cfsai --json build --config .cfs/max78002-csbga.cfsconfig

Sample JSON output

    {"level":"INFO","msg":"Created file \"tensforlfow/src/adi_tflm/hello_world_f32.cpp\"","file_created_event":{"status":"OK","path":"tensforlfow/src/adi_tflm/hello_world_f32.cpp"}}
    {"level":"INFO","msg":"Created file \"tensforlfow/src/adi_tflm/hello_world_f32.hpp\"","file_created_event":{"status":"OK","path":"tensforlfow/src/adi_tflm/hello_world_f32.hpp"}}