Command-Line Interface (CLI)

CLI

eIQ Time Series Studio provides a command-line interface (CLI) in addition to the graphical user interface (GUI).

eIQ Time Series Studio CLI enables you to execute commands quickly and requires fewer computing resources than the graphical interface, which is ideal for to use eIQ Time Series Studio features easily, simply, and concisely by entering text commands.

⚠️ Important Notice: The CLI engine and GUI cannot be used at the same time. Please ensure one is shut down before launching the other.

Install

  1. Go to the eIQ Time Series Studio IDE installation path.

  2. Make sure you can find the following files:

    • tss_cli: The executable file of eIQ Time Series Studio CLI.

    • tss_engine/: The application folder of eIQ Time Series Studio Engine.

Getting started

The following is an overview of the eIQ Time Series Studio Command-Line Interface (TSS CLI). It provides command interfaces for some key features of eIQ Time Series Studio to complete the pipeline from building to deploying a time series model.

image

0. Quick Start of CLI:

Engine

  • Launch or stop the eIQ Time Series Studio Engine. It is mandatory to have a eIQ Time Series Studio Engine running to use eIQ Time Series Studio CLI.
  • Ex: tss_cli engine launch
  • ⚠️Notice: After launching the engine, open a new terminal window to send tss_cli commands.

License

  • Activate your license. ⚠️Notice: It is mandatory to have a license to use the CLI. If you do not have a license key, register in the IDE first to get one.
  • Ex: tss_cli license activate --key MY-LICENSE-KEY

Project

  • Create a BYOD or BYOM project.
  • (BYOD project creation for AD task) Ex: tss_cli project create --project_name my-project --algo_type ad --device FRDM-MCXN947 --channels 3 --label_target_num 2
  • (BYOM project creation for CLS task) Ex: tss_cli project create --project_name my-byom-project --algo_type cls --device FRDM-MCXN947 --channels 3 --label_target_num 2 --conversion true

Signal

  • Import signals.
  • Ex: tss_cli signal import --project_name my-project --signal_name my-signal --file_path /path/to/my/signal --label_id 1 --delimiter " "

Optimization (model training)

  • Start optimization, retrieve progress, and get results for BYOD projects.
  • (CML optimization) Ex: tss_cli optimization start --project_name my-project --opt_name my-opt --signals 1 2 3 4 5 6
  • (DL optimization) Ex: tss_cli optimization start_dl --project_name my-project --opt_name my-dl-opt --signals 1 2 3 4 5 6 --train_model CNN --quant_type INT8
  • Ex: tss_cli optimization query|progress|best|results|suspend|resume|stop|delete --project_name my-project --opt_id 1

Conversion

  • Start model conversion for Bring Your Own Model (BYOM) projects. ⚠️Notice: BYOM projects need to pass in --opt_id=<conversion_id> when using Emulation or Library commands.
  • Ex: tss_cli conversion start --project_name my-byom-project --conversion_name my-conversion --model_path /path/to/my/model --quant_type INT8 --signals 1 2 3 4 5 6
  • Ex: tss_cli conversion query|result|delete --project_name my-byom-project --conversion_id 1

Emulation

  • Launch a Time Series Emulation. (Support multiple model results to emulate simultaneously)
  • (AD/CLS task) Ex: tss_cli emulation launch --project_name my-project --opt_id 1 --result_ids 1 2 3 4 5 --test_file_info /path/to/my/normal/signal "," Normal --test_file_info /path/to/my/anomaly/signal "," Anomaly
  • (REG/OC/FC task) Ex: tss_cli emulation launch --project_name my-project --opt_id 1 --result_ids 1 2 3 4 5 --test_file_info /path/to/my/signal "," " "

Library

  • Compile a Time Series Library.
  • Ex: tss_cli library compile --project_name my-project --opt_id 1 --result_id 1 --save_path /path/to/my/library --arch cortex-m33 --toolchain GCC
Other commands are available, refer to the detailed introduction of each module below.

1. Engine: Launch or shutdown eIQ Time Series Studio Engine

  • Launch the eIQ Time Series Studio Engine on the local computer:

$ tss_cli engine launch --exe_path "</path/to/tss_engine/tss_engine.exe>" --port <port-number> --workspace "</path/to/my/workspace>"
  • Shutdown the eIQ Time Series Studio Engine on the local computer:

$ tss_cli engine shutdown
  • Test connection to the eIQ Time Series Studio Engine:

$ tss_cli engine test
  • Lists all available device information:

$ tss_cli engine device
  • Lists all available sensor information:

$ tss_cli engine sensor
  • Get the number of CPU cores on the running machine:

$ tss_cli engine cpu
  • Get available GPU information on the running machine:

$ tss_cli engine gpu

Usage:

$ tss_cli engine <command> <attributes>

Command:
        launch
                Launch the eIQ Time Series Studio Engine.
                Optional attributes: exe_path, port, workspace
        shutdown
                Shutdown the eIQ Time Series Studio Engine.
        test
                Test connection to the eIQ Time Series Studio Engine.
        device
                Lists all available device information.
        sensor
                Lists all available sensor information.
        cpu
                Get the number of CPU cores on the running machine.
        gpu
                Get available GPU information on the running machine.

Attributes:
        -e, --exe_path file path
                Path to the eIQ Time Series Studio Engine execute file (default `./tss_engine/tss_engine`).
        -p, --port int
                Port of the eIQ Time Series Studio Engine (default 18000).
        -w, --workspace string
                Path to the workspace directory (default `APPDATA/time_series_studio/workspace`).

2. License: activate or deactivate license

  • Activate a license.

    ⚠️Notice: It is mandatory to have a license to use the CLI. If you do not have a license key, register in the IDE first to get one. After successfully logging in to your account, you can click on the account name and then click Copy License to get your license key, which can be used on any platform supported by eIQ Time Series Studio Engine.

    image
$ tss_cli license activate --key "<MY-LICENSE-KEY>"
  • Deactivate the license:

$ tss_cli license deactivate
  • Query current license information:

$ tss_cli license query

Usage:

$ tss_cli license <command> <attributes>

Command:
        activate
                Activate the license.
                Mandatory attributes: key
        deactivate
                Deactivate the license.
        query
                Query current license information.

Attributes:
        -k, --key string
                The license key (If you don't have a license key, register in the IDE first to get one).

Response Info:

Name

Type

Description

Example

email

string

The user’s email address.

“noreply.timeseries@nxp.com”

first_name

string

The user’s first name.

“noreply”

last_name

string

The user’s last name.

“timeseries”

license_key

string

The user’s license key.

“XXXXXX-XXXXXX-XXXXXX-XXXXXX-XXXXXX-XXXXXX”

3. Project: create and manage project

  • List all project information:

$ tss_cli project list
  • Import a project:

$ tss_cli project import --project_zip_path "</path/to/my/project.zip>"
  • Export a project:

$ tss_cli project export --project_name "<my-project>" --save_path "</path/to/save>"
  • Create a BYOD project:

$ tss_cli project create --project_name "<my-project>" --algo_type "<my-algo-type>" --device "<my-deployed-device>" --channels <number-of-channels> --label_target_num <number-of-label-or-target> \
--sensors "<sensor-name>" <count-of-sensor-channels> \
--sensors "<sensor-name>" <count-of-sensor-channels> \
--sensors ...
  • Create a BYOM project:

$ tss_cli project create --project_name "<my-byom-project>" --algo_type "<my-algo-type>" --device "<my-deployed-device>" --channels <number-of-channels> --label_target_num <number-of-label-or-target> --conversion true
  • Create a Forecasting (FC) BYOD project:

$ tss_cli project create --project_name "<my-fc-project>" --algo_type fc --device "<my-deployed-device>" --channels <number-of-channels> \
--fc_channels <target-channel-id> <target-channel-id> \
--fc_recursive <true|false> \
--fc_horizons <horizon> <horizon> \
--fc_cross_instance <true|false>
  • Query project information:

$ tss_cli project query --project_name "<my-project>"
  • Delete a project:

$ tss_cli project delete --project_name "<my-project>"

Usage:

$ tss_cli project <command> <attributes>

Command:
        list
                List all project information.
        import
                Import a project.
                Mandatory attributes: project_zip_path
        export
                Export a project.
                Mandatory attributes: project_name, save_path
        create
                Create a project.
                Mandatory attributes: project_name, algo_type, device, channels
                Mandatory for cls/reg: label_target_num
                Mandatory for fc: fc_channels, fc_recursive, fc_cross_instance
                Mandatory for fc (when fc_recursive is false): fc_horizons
                Note: when fc_recursive is true, fc_horizons is not applicable and will be ignored (fixed to 1 automatically).
                Optional attributes: max_ram, max_flash, description, conversion, sensors
        query
                Query project information.
                Mandatory attributes: project_name
        delete
                Delete a project
                Mandatory attributes: project_name

Attributes:
        -p, --project_name string
                Name of the project.
        -desc, --description string
                Description of the project.
        -a, --algo_type [ad|cls|reg|oc|fc]
                Set algorithm type from [ad, cls, reg, oc, fc], representing anomaly detection, n-Class classification, 1-Class classification, regression and forecasting.
        -d, --device string
                Name of deploy device. You can obtain all available device with command: engine device.
        -mf, --max_flash int
                Maximum Flash available for the library in bytes (default is 0, meaning unlimited).
        -mr, --max_ram int
                Maximum RAM available for the library in bytes (default is 0, meaning unlimited).
        -ch, --channels int
                Number of signal channels.
        -ltn, --label_target_num int
                Number of labels for cls (n-Class classification) or number of targets for reg (regression).
                Required for cls and reg; not applicable for ad, oc or fc.
        -fcc, --fc_channels list
                Target channel ids to forecast among all signal channels (fc only). Required for fc.
                Example: available channels [1,2,3,4,5,6], forecast channels [2,4] means channels 2 and 4 are the target channels whose future values will be predicted.
                Accepts multiple values or ranges in a single flag, e.g. -fcc 1 2 5 or -fcc 1-2 5.
        -fcr, --fc_recursive bool
                Determines the forecasting strategy (fc only). Required for fc.
                true: Enables Recursive Forecasting, where future time steps are predicted sequentially using previous predictions as inputs ('--fc_horizons' is not applicable and will be ignored, horizon fixed to 1).
                false: Uses Direct Multi-Step Forecasting, where all target future time steps are predicted simultaneously in a single output ('--fc_horizons' required).
        -fch, --fc_horizons list
                Specifying which future time steps are predicted by the model in each forecast model output (fc only).
                Example: [1, 3, 6] indicates that the model predicts the values at t+1, t+3, and t+6 directly.
                Accepts multiple values or ranges in a single flag, e.g. -fch 1 3 6 or -fch 1-10.
                Required when '--fc_recursive' / '-fcr' is false. Not applicable when '--fc_recursive' is true (horizon is automatically fixed to 1 and any provided value will be ignored).
        -fci, --fc_cross_instance bool
                Enable cross-instance forecasting (fc only). Required for fc.
                true: The model generalizes across multiple similar instances (e.g., devices); signal import must include data files from multiple instances.
                false: Single-instance mode.
        -cnv, --conversion bool
                Whether to build a Bring Your Own Model (BYOM) project for model conversion. Set to true to enable (default is false).
        -se, --sensors (string, int)
                List of sensors. Input sensor info with (sensor_name, count_of_sensor_channels). You can obtain all available sensor with command: engine sensor.
        -z, --project_zip_path string
                Path to the project zip file (used for import command).
        -svp, --save_path string
                The path to save the project zip file (used for export command).

Response Info:

Name

Type

Description

Example

creation_date

string

The creation date of the project.

“2026-09-15T10:54:16.516712”

name

string

Name of the project.

“my_project”

description

string

Description of the project.

“This is my project.”

algo_type

string

The algorithm type of the project.

“ad”

channels

int

Number of signal channels.

3

channel_name_list

list

The list of channel name.

[“x”,”y”,”z”]

cls_label_num

int

Number of labels of the CLS signals. (cls only)

2

label_id_list

list

The list of label id of the AD/CLS/OC signal. (ad/cls/oc only)

[1,2]

label_name_list

list

The list of label name of the AD/CLS/OC signal. (ad/cls/oc only)

[“Normal”,”Anomaly”]

reg_target_num

int

Number of targets of the REG signals. (reg only)

2

target_info

list

The list of target name of the REG signals. (reg only)

[“1”,”2”]

forecast_channels

list

Target channel ids to forecast among all signal channels. (fc only)

[2,4]

forecast_recursive

bool

Whether recursive forecasting strategy is enabled. (fc only)

false

forecast_horizons

list

The future time steps predicted by the model in each forecast output. (fc only)

[1,3,6]

forecast_cross_instance

bool

Whether cross-instance forecasting is enabled. (fc only)

false

forecast_info

list

Label names of the forecast model outputs. (fc only)

[“Channel_16_5_ahead”]

forecast_num

int

Number of forecast targets. (fc only)

1

max_ram

int

Maximum RAM available for the library, in bytes.

32000

max_flash

int

Maximum Flash available for the library in bytes.

32000

device

string

Name of deployed device.

“FRDM-MCXN947”

device_info

dict

Detailed information of deployed device.

{…}

sensor_info

list

Detailed information of sensor used to collect data.

[…]

conversion

bool

If is a Bring Your Own Model (BYOM) project.

false

summary

dict

Summary of signals and optimizations of the project.

{…}

4. Signal: import signals

  • List all signal information:

$ tss_cli signal list --project_name "<my-project>"
  • Import a signal:

$ tss_cli signal import --project_name "<my-project>" --signal_name "<my-signal>" --file_path "</path/to/my/signal>" --label_id <label-id> --delimiter "<delimiter>"
  • Query a signal information:

$ tss_cli signal query --project_name "<my-project>" --signal_id <signal-id> 
  • Delete a signal:

$ tss_cli signal delete --project_name "<my-project>" --signal_id <signal-id> 
  • Get recommended lookback window sizes (fc tasks only):

$ tss_cli signal recommended_lookback --project_name "<my-fc-project>"

Usage:

$ tss_cli signal <command> <attributes>

Command:
        list
                List all signal information.
                Mandatory attributes: project_name
        import
                Import a signal.
                Mandatory attributes: project_name, signal_name, file_path
                Optional attributes: label_id (required for ad, cls; not applicable for reg, oc, fc), delimiter
        query
                Query signal information.
                Mandatory attributes: project_name, signal_id
        delete
                Delete a signal.
                Mandatory attributes: project_name, signal_id
        recommended_lookback
                Analyse all imported signal files in the project and return the top recommended lookback window sizes. Only applicable for fc tasks.
                Mandatory attributes: project_name

Attributes:
        -p, --project_name string
                Name of the project.
        -sn, --signal_name string
                Name of the signal.
        -si, --signal_id int
                Id of the signal.
        -fp, --file_path file path
                File path to the signal.
        -dlm, --delimiter char
                Signal file value delimiter (default is " ").
        -lid, --label_id int
                Label id of signal. Required for ad [1:Normal, 2:Anomaly] and cls tasks. Not applicable for reg, oc or fc tasks.

Response Info:

Name

Type

Description

Example

name

string

Name of the signal.

“my_signal”

id

int

Id of signal.

1

delimiter

string

Signal file value delimiter.

“,”

file_found

bool

If the signal file has been found on the machine.

true

file_name

string

Signal filename in the workspace.

“1.csv”

info

dict

Signal file information, including file_md5, file_size, and is_raw_data.

{…}

passed

bool

If check on signal data has been successfully passed.

true

label_id

int

Label id of the AD/CLS/OC signal. (ad/cls/oc only)

1

label_name

string

Label name of the AD/CLS/OC signal. (ad/cls/oc only)

“Normal”

target_info

list

The list of target name of the REG signals. (reg only)

[“1”,”2”]

forecast_info

list

Label names of the forecast model outputs. (fc only)

[“Channel_16_5_ahead”]

signal_metadata

dict

Metadata of signal data, including channels, total lines, and values per channel.

{…}

5. Optimization: start optimization, retrieve progress, and get the results for BYOD projects.

  • List all optimization information:

$ tss_cli optimization list --project_name "<my-project>"
  • Start an optimization for classical machine learning:

$ tss_cli optimization start --project_name "<my-project>" --opt_name "<my-opt>"  --signals <signal-id> <signal-id> ... <signal-id> --quick_search true
  • Start an optimization for classical machine learning (fc task):

$ tss_cli optimization start --project_name "<my-fc-project>" --opt_name "<my-opt>" --signals <signal-id> <signal-id> ... <signal-id> --lookback <lookback>
  • Start an optimization for deep learning (not supported for fc projects):

$ tss_cli optimization start_dl --project_name "<my-project>" --opt_name "<my-dl-opt>"  --signals <signal-id> <signal-id> ... <signal-id> --train_model "<train-model>" --quant_type "<quant-type>" --gpu_id <gpu-id>
  • Query the optimization information:

$ tss_cli optimization query --project_name "<my-project>" --opt_id <my-opt-id>
  • Suspend an optimization:

$ tss_cli optimization suspend --project_name "<my-project>" --opt_id <my-opt-id>
  • Resume an optimization:

$ tss_cli optimization resume --project_name "<my-project>" --opt_id <my-opt-id>
  • Stop an optimization:

$ tss_cli optimization stop --project_name "<my-project>" --opt_id <my-opt-id>
  • Get the progression of an optimization:

$ tss_cli optimization progress --project_name "<my-project>" --opt_id <my-opt-id>
  • Get the best result of an optimization:

$ tss_cli optimization best --project_name "<my-project>" --opt_id <my-opt-id>
  • Get all results list of an optimization:

$ tss_cli optimization results --project_name "<my-project>" --opt_id <my-opt-id>
  • Get a result information of the optimization:

$ tss_cli optimization result --project_name "<my-project>" --opt_id <my-opt-id> --result_id <result-id>
  • Delete an optimization:

$ tss_cli optimization delete --project_name "<my-project>" --opt_id <my-opt-id>
  • Query the supported deep learning model types for the project:

$ tss_cli optimization dl_models --project_name "<my-project>"

Usage:

$ tss_cli optimization <command> <attributes>

Command:
        list
                List all optimization information.
                Mandatory attributes: project_name
        start
                Start an optimization for classical machine learning.
                Mandatory attributes: project_name, opt_name, signals
                Optional attributes: on_device_learn, use_generated_anomaly, quick_search, train_val_ratio, allocated_cores
                fc task additional mandatory attributes: lookback
                fc task optional attributes: stride (default 1), recursive_length (required when project fc_recursive=true)
        start_dl
                Start an optimization for deep learning. Not supported for fc (forecasting) projects.
                Mandatory attributes: project_name, opt_name, signals, train_model, quant_type
                Optional attributes: train_val_ratio, allocated_cores, gpu_id
        query
                Query the optimization information.
                Mandatory attributes: project_name, opt_id
        delete
                Delete an optimization.
                Mandatory attributes: project_name, opt_id
        suspend
                Suspend an optimization.
                Mandatory attributes: project_name, opt_id
        resume
                Resume an optimization.
                Mandatory attributes: project_name, opt_id
        stop
                Stop an optimization.
                Mandatory attributes: project_name, opt_id
        progress
                Get the progression of an optimization.
                Mandatory attributes: project_name, opt_id
        best
                Get the best result of an optimization.
                Mandatory attributes: project_name, opt_id
        results
                Get all results list of an optimization.
                Mandatory attributes: project_name, opt_id
        result
                Get a result information of the optimization.
                Mandatory attributes: project_name, opt_id, result_id
        dl_models
                Query the supported deep learning model types for the project.
                Mandatory attributes: project_name

Attributes:
        -p, --project_name string
                Name of the project.
        -on, --opt_name string
                Name of the optimization.
        -sigs, --signals list
                List of signal ids. Accepts multiple values or ranges, e.g. -sigs 1 2 5 or -sigs 1-10.
        -odl, --on_device_learn bool
                Enable on-device learning. Only applicable for ad tasks. Set to true to enable (default is false).
        -uga, --use_generated_anomaly bool
                Enable using generated anomaly data. Only applicable for ad tasks. Set to true to enable (default is false).
        -qs, --quick_search bool
                Enable quick search training. Set to true to enable (default is false).
        -tm, --train_model string
                Name of the deep learning model to train (e.g., CNN). Use 'optimization dl_models' to query supported model types for the project.
        -q, --quant_type string
                Quantization type of the configured deep learning model. Available options: INT8, FP32.
        -r, --train_val_ratio float
                The ratio of train set and val set (default 0.8).
        -lb, --lookback int
                Number of historical time steps used as input for each sample (fc only, required). Use "signal recommended_lookback" to get a suggested value.
        -st, --stride int
                Number of time steps the sliding window advances between samples (fc only, default 1).
        -rl, --recursive_length int
                Number of recursive prediction steps (fc only). Required when the project has fc_recursive=true; ignored otherwise.
        -cores, --allocated_cores int
                Allocated CPU cores. Default: total cores - 4. Use 'engine cpu' to check available cores.
        -gi, --gpu_id int
                GPU ID for model training. Default: -1 (no GPU). Use 'engine gpu' to list available GPUs.
        -oi, --opt_id int
                Id of the optimization. Use 'optimization list' to find the 'id' value for your optimization.
        -ri, --result_id int
                Id of the model result of the optimization. Use 'optimization results' to find the 'id' value for your selected model result.

Response Info:

Name

Type

Description

Example

creation_date

string

The creation date of the optimization.

“2026-09-15T11:07:17.588416”

name

string

Name of the optimization.

“my_opt”

id

int

Id of optimization.

1

signals

list

List of signal data to be used by the optimization.

[…]

max_ram

int

Maximum RAM available for the library in bytes.

32000

max_flash

int

Maximum Flash available for the library in bytes.

32000

state

string

Status of optimization.

“Running”

support_ODL

bool

If on-device learning is enabled for AD task.

false

fast_mode

bool

If quick search training is enabled.

true

use_generated_anomaly

bool

If using generated anomaly data is enabled for AD task.

false

shuffling

bool

If training data shuffling is enabled.

false

train_val_ratio

float

The ratio of train set and val set.

0.8

cpu_info

dict

CPU core usage of the optimization.

{…}

progress

dict

The progress of the optimization.

{…}

channel_name_list

list

The list of channel name.

[“x”,”y”,”z”]

label_id_list

list

The list of label id of the AD/CLS/OC signal. (ad/cls/oc only)

[1,2,5]

label_name_list

list

The list of label name of the AD/CLS/OC signal. (ad/cls/oc only)

[“Normal”,”Anomaly”]

target_info

list

The list of target name of the REG signals. (reg only)

[“1”,”2”]

forecast_channels

list

Target channel ids to forecast among all signal channels. (fc only)

[16]

forecast_recursive

bool

Whether recursive forecasting strategy is enabled. (fc only)

false

forecast_horizons

list

The future time steps predicted by the model in each output. (fc only)

[5]

forecast_cross_instance

bool

Whether cross-instance forecasting is enabled. (fc only)

false

forecast_info

list

Label names of the forecast model outputs. (fc only)

[“Channel_16_5_ahead”]

lookback

int

Number of historical time steps used as input per sample. (fc only)

1

stride

int

Sliding window step size between samples. (fc only)

1

recursive_length

int

Number of recursive prediction steps. (fc only, when fc_recursive=true)

1

threshold

float

Decision threshold used by the model. (ad/cls only)

0.0

model_type

string

The type of training models.

“CML”

quant_type

string

Quantization type of the configured deep learning model.

“INT8”

configured_models_info

list

The information of configured deep learning models.

[…]

version

string

The version number of the eIQ Time Series Studio.

“2.1.3”

  • optimization results:

Name

Type

Description

Example

id

int

Id of the model result of the optimization.

190

is_best

bool

If is best in the optimization.

true

channels

int

Number of signal channels.

16

nb_cols

int

Number of signal columns (input features).

16

sp_info

dict

Signal processing pipeline applied to the model input.

{…}

model_info

dict

Model information including name, description, type, task and enable status.

{…}

benchmark

dict

Benchmark results including performance metrics, memory usage, and execution time.

{…}

validation_result

dict

Validation set predictions.

{…}

Emulation

list

The emulation results list of the model.

[…]

6. Conversion: Start model conversion for Bring Your Own Model (BYOM) projects.

⚠️ Notice: BYOM projects need to pass in –opt_id=<conversion_id> when using Emulation or Library commands.

  • List all conversion information:

$ tss_cli conversion list --project_name "<my-byom-project>"
  • Start a model conversion:

$ tss_cli conversion start --project_name "<my-byom-project>" --conversion_name "<my-conversion>" --model_path "</path/to/my/model>" --quant_type "<quant-type>" --signals <signal-id> <signal-id> ... <signal-id>
  • Query the conversion information:

$ tss_cli conversion query --project_name "<my-byom-project>" --conversion_id <my-conversion-id>
  • Get the converted model results:

$ tss_cli conversion result --project_name "<my-byom-project>" --conversion_id <my-conversion-id>
  • Delete a conversion:

$ tss_cli conversion delete --project_name "<my-byom-project>" --conversion_id <my-conversion-id>

Usage:

$ tss_cli conversion <command> <attributes>

Command:
        list
                List all conversion information.
                Mandatory attributes: project_name
        start
                Start a model conversion.
                Mandatory attributes: project_name, conversion_name, model_path, quant_type, signals
        query
                Query the conversion information.
                Mandatory attributes: project_name, conversion_id
        delete
                Delete a conversion.
                Mandatory attributes: project_name, conversion_id
        result
                Get the converted model results.
                Mandatory attributes: project_name, conversion_id

Attributes:
        -p, --project_name string
                Name of the project.
        -cn, --conversion_name string
                Name of the conversion.
        -ci, --conversion_id int
                Id of the conversion.
        -mp, --model_path string
                Path to the model file.
        -q, --quant_type [INT8|FP32]
                Quantization type of the model.
        -sigs, --signals list
                List of signal ids. Accepts multiple values or ranges, e.g. -sigs 1 2 5 or -sigs 1-10.

Response Info:

Name

Type

Description

Example

creation_date

string

The creation date of the conversion.

“Mon, 25 Aug 2025 15:51:33 GMT”

name

string

Name of the conversion.

“my_conversion”

id

int

Id of conversion.

1

signals

list

List of signal data to be used by the conversion.

[…]

max_ram

int

Maximum RAM available for the library in bytes.

32000

max_flash

int

Maximum Flash is available for the library in bytes.

32000

model_path

string

Path to the model file.

“/path/to/my/model”

label_id_list

list

The list of label id of the AD/CLS/OC signal.

[1,2,5]

label_name_list

list

The list of label name of the AD/CLS/OC signal.

[“1”,”2”,”5”]

target_info

list

The list of target name of the REG signals.

[“1”,”2”]

quant_type

string

Quantization type of the model conversion.

“INT8”

version

string

The version number of the eIQ Time Series Studio.

“99.99.0”

7. Emulation: launch time series emulation of multiple model results.

  • Launch an emulation (supports multiple model results simultaneously):

$ tss_cli emulation launch --project_name "<my-project>" --opt_id <my-opt-id> --result_ids <result-id> <result-id> ... <result-id> \
--test_file_info "</path/to/my/test/signal>" "<delimiter>" "<label-name>" \
--test_file_info "</path/to/my/test/signal>" "<delimiter>" "<label-name>" \
--test_file_info ...

Note on label_name:

  • For ad tasks: label_name must be one of Normal, Anomaly, or learn (learn is only valid when the optimization was trained with on_device_learn=true).

  • For cls tasks: label_name must be chosen from the label_name_list of the optimization (use optimization query to view available label names).

  • For other task types (reg, oc, fc): label_name is not used — pass any placeholder, e.g. " ".

  • Retry on an emulation model:

$ tss_cli emulation retry --project_name "<my-project>" --emulation_id <my-emulation-id> --result_id <my-result-id>
  • Delete an emulation:

$ tss_cli emulation delete --project_name "<my-project>" --emulation_id <my-emulation-id>
  • Start an emulation for unlabeled test files and return prediction result files (not supported for fc projects):

$ tss_cli emulation predict --project_name "<my-project>" --opt_id <my-opt-id> --result_id <my-result-id> \
--unlabeled_test_file_info "</path/to/my/test/signal>" "<delimiter>" \
--unlabeled_test_file_info "</path/to/my/test/signal>" "<delimiter>" \
--unlabeled_test_file_info ...

Usage:

$ tss_cli emulation <command> <attributes>

Command:
        launch
                Launch an emulation.(Support multiple model results to emulate simultaneously)
                Mandatory attributes: project_name, opt_id, result_ids, test_file_info
        retry
                Retry on an emulation model.
                Mandatory attributes: project_name, emulation_id, result_id
        delete
                Delete an emulation.
                Mandatory attributes: project_name, emulation_id
        predict
                Start an emulation for unlabeled test files and return prediction result files. Not supported for fc (forecasting) projects.
                Mandatory attributes: project_name, opt_id, result_id, unlabeled_test_file_info

Attributes:
        -p, --project_name string
                Name of the project.
        -oi, --opt_id int
                Id of the optimization. Use 'optimization list' to find the 'id' value for your optimization.
        -ri, --result_id int
                Id of the model result that need to be emulated. Use 'optimization results' to find the 'id' value for your selected model result.
        -ris, --result_ids list
                List of result_ids of the models that need to be emulated. Accepts multiple values or ranges, e.g. -ris 1 2 5 or -ris 1-10.
        -t, --test_file_info (str, str, str)
                List of test files. Input test file with (file_path, delimiter, label_name).
                For ad tasks: label_name must be one of [Normal, Anomaly, learn] (learn is only valid when the optimization was trained with on_device_learn=true).
                For cls tasks: label_name must be chosen from the label_name_list of the optimization (use 'optimization query' to view available label names).
                For other task types: label_name is not used (pass any placeholder, e.g. " ").
        -ut, --unlabeled_test_file_info (str, str)
                List of unlabeled test files. Input test file with (file_path, delimiter).
        -ei, --emulation_id int
                Id of the emulation.

Response Info:

Name

Type

Description

Example

creation_date

string

The creation date of the emulation.

“Mon, 09 Dec 2024 11:42:03 GMT”

emulation_id

int

Id of the emulation.

1

emulation_results

list

The list of emulation results for each model result.

[…]

  • emulation results:

Name

Type

Description

Example

result_id

int

Id of the model result of the optimization.

18

status

string

The status of the emulation for the model result.

“Success”

file_performance

dict

The performance of the model result for each test file.

{…}

label_performance

dict

The performance of the model result for each label in AD/CLS tasks.

{…}

total_performance

dict

The total performance of the model result.

{…}

8. Library: Generates the time series library or sample project.

  • Compile a Time Series Library for an optimization:

$ tss_cli library compile --project_name "<my-project>" --opt_id <my-opt-id>  --result_id <my-result-id> --save_path "</path/to/my/library>" --arch "<arch-type>" --toolchain "<my-toolchain>" --chip "<my-chip>" --fshort_enums true --powerquad_enable true
  • Get the sample project for a library:

$ tss_cli library sample_project --project_name "<my-project>" --opt_id <my-opt-id>  --result_id <my-result-id> --save_path "</path/to/my/project>" --arch "<arch-type>" --toolchain "<my-toolchain>" --chip "<my-chip>" --neutron_enable true
  • Estimate the inference time of a library:

$ tss_cli library time_estimate --project_name "<my-project>" --opt_id <my-opt-id>  --result_id <my-result-id>

Usage:

$ tss_cli library <command> <attributes>

Command:
        compile
                Compile a Time Series Library for an optimization.
                Mandatory attributes: project_name, opt_id, result_id, save_path, arch, toolchain
                Optional attributes: save_name, chip, mfloat_abi, fshort_enums, fshort_wchar, powerquad_enable, neutron_enable
        sample_project
                Get the sample project for a library.
                Mandatory attributes: project_name, opt_id, result_id, save_path, arch, toolchain
                Optional attributes: save_name, chip, mfloat_abi, fshort_enums, fshort_wchar, powerquad_enable, neutron_enable
        time_estimate
                Estimate the inference time of a library.
                Mandatory attributes: project_name, opt_id, result_id

Attributes:
        -p, --project_name string
                Name of the project.
        -oi, --opt_id int
                Id of the optimization.
        -ri, --result_id int
                Id of the model result of the optimization.
        -svp, --save_path string
                The path to the library or sample_project to save.
        -svn, --save_name string
                The name of the library or sample_project to save (default `libtss` for compile, `protss` for sample_project).
        -ar, --arch [host|cortex-m7|cortex-m4|cortex-m33|cortex-m0plus|DSP56800EF|DSP56800EX]
                Set arch type.
        -cp, --chip string
                Chip part number (default `MCXN947`).
        -t, --toolchain [GCC|IAR|Keil|CodeWarrior]
                Set toolchain type.
        -fabi, --mfloat_abi [soft|hard]
                Select floating-point ABI to use (default `hard`).
        -fse, --fshort_enums bool
                Enable fshort-enums (default false).
        -fsw, --fshort_wchar bool
                Enable fshort-wchar (default false).
        -pq, --powerquad_enable bool
                Enable the PowerQuad accelerator (default false).
        -nt, --neutron_enable bool
                Enable the Neutron accelerator (default false).