diff --git a/README.md b/README.md
index 08617aa..d53e744 100644
--- a/README.md
+++ b/README.md
@@ -2,7 +2,7 @@
 
 [![Lint and test](https://github.com/uhh-cms/MLProf/actions/workflows/lint.yml/badge.svg)](https://github.com/uhh-cms/MLProf/actions/workflows/lint.yml)
 
-Tools for automated ML model performance tests in CMSSW
+Tools for automated ML model performance tests in CMSSW (CMSSW version 13 and above).
 
 ## Quickstart
 
@@ -46,7 +46,7 @@ flowchart TD
 ```
 
 A task is run with the command `law run` followed by the name of the task.
-A version, given by the argument `--version```, followed by the name of the version, is required.
+A version, given by the argument `--version`, followed by the name of the version, is required.
 Each `version` has its own set of outputs for the different existing tasks.
 
 In law, the intermediate results (=the outputs to the different tasks) are saved locally in the corresponding directory (default path in MLPROF is ```data/name_of_the_task/CMSSW_architecture/model_identifier_with_run_parameters/name_of_the_version/```).
@@ -56,7 +56,7 @@ Tasks in law are organized as a graph with dependencies.
 Therefore a "depth" for the different required tasks exists, depending on which task required which other task.
 In order to see the different required tasks for a single task, you might use the argument `--print-status -1`, which will show all required tasks and the existence or absence of their output for the given input parameters up to depth "-1", hence the deepest one.
 The called task with `law run` will have depth 0.
-You might check the output path of a task with the argument `--print-output```, followed by the depth of the task.
+You might check the output path of a task with the argument `--print-output`, followed by the depth of the task.
 If you want a finished task to be run anew without changing the version (e.g. do a new runtime measurement with a new training of the same original network), you might remove the previous outputs with the `--remove-output` argument, followed by the depth up to which to remove the outputs.
 There are three removal modes:
 - `a` (all: remove all outputs of the different tasks up to the given depth),
@@ -98,29 +98,23 @@ flowchart TD
     A[CreateRuntimeConfig] --> B[MeasureRuntime]
     B --> |merge the results for different batch sizes| C[MergeRuntimes]
     C --> D[PlotRuntimes]
-    C --> E[PlotRuntimesMultipleCMSSW]
-    C --> F[PlotRuntimesMultipleNetworks]
+    C --> E[PlotRuntimesMultipleParams]
 ```
 
 It is composed of four major types of tasks:
 
-1. [CreateRuntimeConfig](#createruntimeconfig): This task creates the cmssw config file to run the inference, using a
-json file for the model parameters.
+1. [CreateRuntimeConfig](#createruntimeconfig): This task creates the cmssw config file to run the inference, using a json file for the model parameters.
 
-2. [MeasureRuntime](#measureruntime): This task runs the network as many times as demanded in the arguments for a
-single batch size and outputs a .csv file with the results of the timing measurements.
+2. [MeasureRuntime](#measureruntime): This task runs the network as many times as demanded in the arguments for a single batch size and outputs a .csv file with the results of the timing measurements.
 
-3. [MergeRuntimes](#mergeruntimes): This task merges the .csv output files with the required multiple batch sizes
-from the [MeasureRuntime](#measureruntime) tasks to obtain a single .csv file containing the informations to plot.
+3. [MergeRuntimes](#mergeruntimes): This task merges the .csv output files with the required multiple batch sizes from the [MeasureRuntime](#measureruntime) tasks to obtain a single .csv file containing the informations to plot.
 
-4. [PlotRuntimes](#plotruntimes), [PlotRuntimesMultipleNetworks](#plotruntimesseveralnetworks),
-[PlotRuntimesMultipleCMSSW](#plotruntimesmultiplecmssw): These tasks create the plots with the values stored in the
-.csv file from [MergeRuntimes](#mergeruntimes).
+4. [PlotRuntimes](#plotruntimes), [PlotRuntimesMultipleParams](#plotruntimesmultipleparams): These tasks create the plots with the values stored in the .csv file from [MergeRuntimes](#mergeruntimes).
 
 Calling the [PlotRuntimes](#plotruntimes) task triggers the whole pipeline with the correct arguments.
 
 The way to give the necessary informations about your model to MLProf is by using a json file.
-Its structure is presented below in [Model file in json format](#model-file-in-json-format)
+Its structure is presented below in [Model file in json format](#model-file-in-json-format).
 
 # Model file in json format
 
@@ -128,7 +122,7 @@ The format of the file to give to MLProf is the following:
 
 ```json
 {
-    "file": "{absolute_path_to_you_pb_model_file}",
+    "file": "{path_to_your_pb_model_file}",
     "inputs": [
         {
             "name": "{name_input_layer_1}",
@@ -149,35 +143,34 @@ The format of the file to give to MLProf is the following:
         },
         ...
     ],
-    "network_name": "{name_of_the_network_for_the_legend_of_the_plots_and_the_name_of_the_output_pdf_of_PlotRuntimesMultipleNetworks}"
+    "inference_engine": "{name_of_inference_engine_(either_tf_or_onnx)}",
+    "network_name": "{optional_name_of_the_network_for_the_labels}"
 }
 ```
-There are already a few examples of these configswith working paths for the networks in the "examples" folder.
+There are already a few examples of these configs with working paths for the networks in the "examples" folder.
 
 # CreateRuntimeConfig
 
-This task create the CMSSW config file to run the inference in the corresponding task, using the template file in
-the `cmssw/MLProf/RuntimeModule/test/` directory.
-The parameters of the inference except the batch sizes are fixed by the created
-configuration file, therefore this task will be run again for every change in the inference (e.g. the number of runs for the statistics, the path to the graph to check...).
+This task create the CMSSW config file to run the inference in the corresponding task, using the template file in the `cmssw/MLProf/RuntimeModule/test/` directory.
+The parameters of the inference except the batch sizes are fixed by the created configuration file, therefore this task will be run again for every change in the inference (e.g. the number of runs for the statistics, the path to the graph to check...).
 
 ## Parameters:
 
-- model-file: str. The absolute path of the json file containing the informations of the model to be tested. default: `$MLP_BASE/examples/model1/model.json`.
+- model-file: str. The absolute path of the json file containing the informations of the model to be tested. default: `$MLP_BASE/examples/simple_dnn/model.json`.
 
-- model-name: str. When set, use this name for the path used when storing outputs instead of a hashed version of `--model-file```. default: empty.
+- model-name: str. When set, use this name for the path used when storing outputs instead of a hashed version of `--model-file`. default: empty.
 
-- input-files: str. Comma-separated list of absolute paths of input files for the CMSSW analyzer (TODO: not implemented); when empty, random input values will be used. default: empty.
+- model-label: str. When set, use this string for the model label in the plots from the plotting tasks. When empty, the `network_name` field in the model json data is used when existing, and model-name otherwise. default: empty.
 
-- events: int. The number of events to read from each input file for averaging measurements. default: `1`.
+- n-events: int. The number of events to read from each input file for averaging measurements. default: `1`.
 
-- repetitions: int. The number of repetitions to be performed per evaluation for averaging. default: `100`.
+- n-calls: int. The number of calls to be performed per evaluation for averaging. default: `100`.
 
-- warmup: int. The number of evaluations to be performed before starting the actual measurement. default: `10`.
+- input-type: str. Define the kind of inputs used during the measurement, either 'random', 'incremental', 'zeros', or a path to a root file. default: random
 
-- cmssw-version: str. The CMSSW version used for the inference. default: `CMSSW_12_2_4`.
+- cmssw-version: str. The CMSSW version used for the inference. default: `CMSSW_13_3_1`.
 
-- scram-arch: str. The SCRAM architecture used for the inference. default: `slc7_amd64_gcc10`.
+- scram-arch: str. The SCRAM architecture used for the inference. default: `slc7_amd64_gcc12`.
 
 ## Output:
 
@@ -188,17 +181,17 @@ configuration file, therefore this task will be run again for every change in th
 ```shell
 law run CreateRuntimeConfig \
     --version test_simple_dnn \
-    --model-file "$MLP_BASE/examples/model1/model.json" \
+    --model-file "$MLP_BASE/examples/simple_dnn/model.json" \
     --model-name dnn \
-    --repetitions 500 \
-    --cmssw-version CMSSW_12_2_4
+    --n-calls 500 \
+    --cmssw-version CMSSW_13_3_1
 ```
 
 
 # MeasureRuntime
 
 Task to provide the time measurements of the inference of a network in CMSSW, given the input parameters and a single batch size.
-The batch size and the (```repetitions * events```) measured values in milliseconds are saved in csv format.
+The batch size and the (```n-calls * n-events```) measured values in milliseconds are saved in csv format.
 
 ## Requires:
 
@@ -208,265 +201,215 @@ The batch size and the (```repetitions * events```) measured values in milliseco
 
 - batch-size: int. the batch size to measure the runtime for. default: `1`.
 
-- model-file: str. The absolute path of the json file containing the informations of the model to be tested. default: `$MLP_BASE/examples/model1/model.json`.
+- model-file: str. The absolute path of the json file containing the informations of the model to be tested. default: `$MLP_BASE/examples/simple_dnn/model.json`.
 
 - model-name: str. When set, use this name for the path used when storing outputs instead of a hashed version of `--model-file`. default: empty.
 
-- input-files: str. comma-separated list of absolute paths of input files for the CMSSW analyzer (TODO: not implemented). When empty, random input values will be used. default: empty.
+- model-label: str. When set, use this string for the model label in the plots from the plotting tasks. When empty, the `network_name` field in the model json data is used when existing, and model-name otherwise. default: empty.
 
-- events: int. The number of events to read from each input file for averaging measurements. default: `1`.
+- n-events: int. The number of events to read from each input file for averaging measurements. default: `1`.
 
-- repetitions: int. The number of repetitions to be performed per evaluation for averaging. default: `100`.
+- n-calls: int. The number of calls to be performed per evaluation for averaging. default: `100`.
 
-- warmup: int. The number of evaluations to be performed before starting the actual measurement. default: `10`.
+- input-type: str. Define the kind of inputs used during the measurement, either 'random', 'incremental', 'zeros', or a path to a root file. default: random
 
-- cmssw-version: str. The CMSSW version used for the inference. default: `CMSSW_12_2_4`.
+- cmssw-version: str. The CMSSW version used for the inference. default: `CMSSW_13_3_1`.
 
-- scram-arch: str. The SCRAM architecture used for the inference. default: `slc7_amd64_gcc10`.
+- scram-arch: str. The SCRAM architecture used for the inference. default: `slc7_amd64_gcc12`.
 
 ## Output:
-- `runtime_bs_{batch-size}.csv```: The batch size and measured values of the runtime
+- `runtime_bs_{batch-size}.csv`: The batch size and measured values of the runtime
 for each repetition and event.
 
 ## Example:
 
 ```shell
 law run MeasureRuntime --version test_simple_dnn \
-                       --model-file $MLP_BASE/examples/model1/model.json \
+                       --model-file $MLP_BASE/examples/simple_dnn/model.json \
                        --model-name dnn \
-                       --repetitions 500 \
-                       --cmssw-version CMSSW_12_2_4 \
+                       --n-calls 500 \
+                       --cmssw-version CMSSW_13_3_1 \
                        --batch-size 1
 ```
 
 
 # MergeRuntimes
 
-This task merges the .csv output files with the required multiple batch sizes
-from the different occurences of the `MeasureRuntime` task to obtain a single .csv
-file containing the informations to plot.
+This task merges the .csv output files with the required multiple batch sizes from the different occurences of the `MeasureRuntime` task to obtain a single .csv file containing the informations to plot.
 
 ## Requires:
 - The .csv files from the several occurences of `MeasureRuntime` (one for each batch size).
 
 ## Parameters:
-- batch-sizes: int. The comma-separated list of batch sizes to be tested; default: `1,2,4```.
+- batch-sizes: int. The comma-separated list of batch sizes to be tested; default: `1,2,4`.
 
-- model-file: str. The absolute path of the json file containing the informations of the model to be tested.
-                   default: `$MLP_BASE/examples/model1/model.json```.
+- model-file: str. The absolute path of the json file containing the informations of the model to be tested. default: `$MLP_BASE/examples/simple_dnn/model.json`.
 
-- model-name: str. When set, use this name for the path used when storing outputs instead of a hashed version of
-        `--model-file```. default: empty.
+- model-name: str. When set, use this name for the path used when storing outputs instead of a hashed version of `--model-file`. default: empty.
 
-- input-files: str. comma-separated list of absolute paths of input files for the CMSSW analyzer (TODO: not implemented);
-                    when empty, random input values will be used; default: empty
+- model-label: str. When set, use this string for the model label in the plots from the plotting tasks. When empty, the `network_name` field in the model json data is used when existing, and model-name otherwise. default: empty.
 
-- events: int. The number of events to read from each input file for averaging measurements. default: `1```
+- n-events: int. The number of events to read from each input file for averaging measurements. default: `1`.
 
-- repetitions: int. The number of repetitions to be performed per evaluation for averaging. default: `100```
+- n-calls: int. The number of calls to be performed per evaluation for averaging. default: `100`.
 
-- warmup: int. The number of evaluations to be performed before starting the actual measurement.
-                    default: `10```
+- input-type: str. Define the kind of inputs used during the measurement, either 'random', 'incremental', 'zeros', or a path to a root file. default: random
 
-- cmssw-version: str. The CMSSW version used for the inference. default: `CMSSW_12_2_4```
+- cmssw-version: str. The CMSSW version used for the inference. default: `CMSSW_13_3_1`
 
-- scram-arch: str. The SCRAM architecture used for the inference. default: `slc7_amd64_gcc10```
+- scram-arch: str. The SCRAM architecture used for the inference. default: `slc7_amd64_gcc12`
 
 ## Output:
-- `runtime_bs_{batch_size_1}_{batch_size_2}_{...}.csv```: The batch size and measured values of the runtime
-for each repetition and event in the several measurements.
+- `runtime_bs_{batch_size_1}_{batch_size_2}_{...}.csv`: The batch size and measured values of the runtime for each repetition and event in the several measurements.
 
 ## Example:
 
 ```shell
 law run MergeRuntimes --version test_simple_dnn \
-                      --model-file $MLP_BASE/examples/model1/model.json \
+                      --model-file $MLP_BASE/examples/simple_dnn/model.json \
                       --model-name dnn \
-                      --repetitions 500 \
-                      --cmssw-version CMSSW_12_2_4 \
+                      --n-calls 500 \
+                      --cmssw-version CMSSW_13_3_1 \
                       --batch-sizes 1,2,4,8,16,32,64,128,256,512,1024
 ```
 
 # PlotRuntimes
 
-This task plots the results of the runtime measurement against the given batch sizes. The points are
-given by the median of the data series and the boundaries of the uncertainty bands are given by the 16 and
-84 percentiles of the data series (Therefore the uncertainty band contains 68% of the data points,
-which corresponds to a $1\sigma$ uncertainty for gaussian uncertainties). The number of inferences behind one
-plotted data point is given by `events * repetitions```.
+This task plots the results of the runtime measurement against the given batch sizes.
+The points are given by the median of the data series and the boundaries of the uncertainty bands are given by the 16 and 84 percentiles of the data series (Therefore the uncertainty band contains 68% of the data points, which corresponds to a $1\sigma$ uncertainty for gaussian uncertainties).
+The number of inferences behind one plotted data point is given by `n-events * n-calls`.
 
 ## Requires:
 - The .csv file from the `MergeRuntimes` task.
 
 ## Parameters:
-- log-y: bool. Plot the y-axis values logarithmically; default: `False```.
+- log-y: bool. Plot the y-axis values logarithmically; default: `False`.
+
+- bs-normalized: bool. Normalize the measured values with the batch size before plotting; default: `True`.
 
-- bs-normalized: bool. Normalize the measured values with the batch size before plotting; default: `True```.
+- filling: bool. Plot the errors as error bands instead of error bars; default: `True`.
 
-- filling: bool. Plot the errors as error bands instead of error bars; default: `True```.
+- top-right-label: str. When set, stick this string as label over the top right corner of the plot. default: empty.
+
+- batch-sizes: int. The comma-separated list of batch sizes to be tested; default: `1,2,4`.
+
+- model-file: str. The absolute path of the json file containing the informations of the model to be tested. default: `$MLP_BASE/examples/simple_dnn/model.json`.
+
+- model-name: str. When set, use this name for the path used when storing outputs instead of a hashed version of `--model-file`. default: empty.
 
-- batch-sizes: int. The comma-separated list of batch sizes to be tested; default: `1,2,4```.
+- model-label: str. When set, use this string for the model label in the plots from the plotting tasks. When empty, the `network_name` field in the model json data is used when existing, and model-name otherwise. default: empty.
 
-- model-file: str. The absolute path of the json file containing the informations of the model to be tested.
-                   default: `$MLP_BASE/examples/model1/model.json```.
+- n-events: int. The number of events to read from each input file for averaging measurements. default: `1`.
 
-- model-name: str. When set, use this name for the path used when storing outputs instead of a hashed version of
-        `--model-file```. default: empty.
+- n-calls: int. The number of calls to be performed per evaluation for averaging. default: `100`.
 
-- input-files: str. comma-separated list of absolute paths of input files for the CMSSW analyzer (TODO: not implemented);
-                    when empty, random input values will be used; default: empty
+- input-type: str. Define the kind of inputs used during the measurement, either 'random', 'incremental', 'zeros', or a path to a root file. default: random
 
-- events: int. The number of events to read from each input file for averaging measurements. default: `1```
+- cmssw-version: str. The CMSSW version used for the inference. default: `CMSSW_13_3_1`
 
-- repetitions: int. The number of repetitions to be performed per evaluation for averaging. default: `100```
+- scram-arch: str. The SCRAM architecture used for the inference. default: `slc7_amd64_gcc12`
 
-- warmup: int. The number of evaluations to be performed before starting the actual measurement.
-                    default: `10```
+- file_types: str. Comma-separated types of the output plot files. default: "pdf".
+
+- plot_postfix: str. An arbitrary postfix that is added with two underscores to all paths of produced plots. default: empty.
+
+- view_cmd: str. A command to execute after the task has run to visualize plots right in the terminal (e.g. "imgcat" if installed). default: empty.
 
-- cmssw-version: str. The CMSSW version used for the inference. default: `CMSSW_12_2_4```
 
-- scram-arch: str. The SCRAM architecture used for the inference. default: `slc7_amd64_gcc10```
 
 ## Output:
-- `runtime_plot_different_batch_sizes_{batch_size_1}_{batch_size_2}_{...}.pdf```: The plot of the runtime measurement
-against the different batch sizes given.
+- `runtime_plot_different_batch_sizes_{batch_size_1}_{batch_size_2}_{...}.pdf`: The plot of the runtime measurement against the different batch sizes given.
 
 ## Example:
 
 ```shell
 law run PlotRuntimes --version test_simple_dnn \
-                     --model-file $MLP_BASE/examples/model1/model.json \
+                     --model-file $MLP_BASE/examples/simple_dnn/model.json \
                      --model-name dnn \
-                     --repetitions 500 \
-                     --cmssw-version CMSSW_12_2_4 \
+                     --model-label "dnn with CMSSW_13_3_1" \
+                     --n-calls 500 \
+                     --cmssw-version CMSSW_13_3_1 \
                      --batch-sizes 1,2,4,8,16,32,64,128,256,512,1024 \
                      --log-y False \
                      --bs-normalized True
 ```
 
-# PlotRuntimesMultipleNetworks
+# PlotRuntimesMultipleParams
 
-This task plots the results of the runtime measurement against the given batch sizes for several models.
-The model-files argument is required and replaces the module-file argument. The points are
-given by the median of the data series and the boundaries of the uncertainty bands are given by the 16 and
-84 percentiles of the data series (Therefore the uncertainty band contains 68% of the data points,
-which corresponds to a $1\sigma$ uncertainty for gaussian uncertainties). The number of inferences behind one
-plotted data point is given by `events * repetitions```.
+This task plots the results of the runtime measurement against the given batch sizes for several values of the given parameters.
+The model-files argument is required and replaces the module-file argument.
+The cmssw-versions argument replaces the cmssw-version arguments, but both have the same default.
+The scram-archs argument replaces the scram-arch argument, but both have the same default.
+These arguments accept brace expansions, see below in the Example subsection.
+The model-names and model-labels arguments replace repectively the model-name and model-label arguments and are empty by default.
+The points are given by the median of the data series and the boundaries of the uncertainty bands are given by the 16 and 84 percentiles of the data series (Therefore the uncertainty band contains 68% of the data points, which corresponds to a $1\sigma$ uncertainty for gaussian uncertainties).
+The number of inferences behind one plotted data point is given by `n-events * n-calls`.
 
 ## Requires:
-- The .csv file from the `MergeRuntimes` task.
+- The .csv files from the `MergeRuntimes` task.
 
 ## Parameters:
-- model-files: str. The comma-separated list of the absolute paths of the json files containing the
-                    informations of the model to be tested. No default value.
-
-- log-y: bool. Plot the y-axis values logarithmically; default: `False```.
-
-- bs-normalized: bool. Normalize the measured values with the batch size before plotting; default: `True```.
-
-- filling: bool. Plot the errors as error bands instead of error bars; default: `True```.
-
-- batch-sizes: int. The comma-separated list of batch sizes to be tested; default: `1,2,4```.
-
-- model-name: str. When set, use this name for the path used when storing outputs instead of a hashed version of
-        `--model-file```. default: empty.
+- model-files: str. The comma-separated list of the absolute paths of the json files containing the informations of the model to be tested. No default value.
 
-- input-files: str. comma-separated list of absolute paths of input files for the CMSSW analyzer (TODO: not implemented);
-                    when empty, random input values will be used; default: empty
+- cmssw-versions: str. The comma-separated list of CMSSW versions used for the inference. default: `CMSSW_13_3_1`
 
-- events: int. The number of events to read from each input file for averaging measurements. default: `1```
+- scram-archs: str. The comma-separated list of SCRAM architectures used for the inference. default: `slc7_amd64_gcc12`
 
-- repetitions: int. The number of repetitions to be performed per evaluation for averaging. default: `100```
+- model-names: str. The comma-separated list of model names. When set, use these names for the path used when storing outputs instead of a hashed version of `--model-file`. default: empty.
 
-- warmup: int. The number of evaluations to be performed before starting the actual measurement.
-                    default: `10```
+- model-labels: str. The comma-separated list of model labels. When set, use these strings for the model labels in the plots from the plotting tasks. When empty, the `network_name` fields in the models json data is used when existing, and model-names otherwise. default: empty.
 
-- cmssw-version: str. The CMSSW version used for the inference. default: `CMSSW_12_2_4```
+- log-y: bool. Plot the y-axis values logarithmically; default: `False`.
 
-- scram-arch: str. The SCRAM architecture used for the inference. default: `slc7_amd64_gcc10```
+- bs-normalized: bool. Normalize the measured values with the batch size before plotting; default: `True`.
 
-## Output:
-- `runtime_plot_networks_{network_name_1}_{network_name_2}_{...}_different_batch_sizes_{batch_size_1}_{batch_size_2}_{...}.pdf```: The plot of the runtime measurement
-against the different batch sizes given.
-
-## Example:
+- top-right-label: str. When set, stick this string as label over the top right corner of the plot. default: empty.
 
-```shell
-law run PlotRuntimesMultipleNetworks --version test_several_networks \
-                                     --model-files $MLP_BASE/examples/model1/model.json,$MLP_BASE/examples/cnn/model_cnn.json\
-                                     --repetitions 500 \
-                                     --cmssw-version CMSSW_12_2_4 \
-                                     --batch-sizes 1,2,4,8,16,32,64,128,256,512,1024 \
-                                     --log-y False \
-                                     --bs-normalized True
-```
+- filling: bool. Plot the errors as error bands instead of error bars; default: `True`.
 
-# PlotRuntimesMultipleCMSSW
+- batch-sizes: int. The comma-separated list of batch sizes to be tested; default: `1,2,4`.
 
-This task plots the results of the runtime measurement against the given batch sizes for several CMSSW versions.
-The cmssw-versions argument replaces the cmssw-version argument. The cmssw-versions argument accepts brace expansions,
-see below in the Example subsection. The points are
-given by the median of the data series and the boundaries of the uncertainty bands are given by the 16 and
-84 percentiles of the data series (Therefore the uncertainty band contains 68% of the data points,
-which corresponds to a $1\sigma$ uncertainty for gaussian uncertainties). The number of inferences behind one
-plotted data point is given by `events * repetitions```.
+- n-events: int. The number of events to read from each input file for averaging measurements. default: `1`
 
-## Requires:
-- The .csv file from the `MergeRuntimes` task.
-
-## Parameters:
-- cmssw-versions: str. The comma separated list of CMSSW version used for the inference.
-                       default: `"CMSSW_12_2_4","CMSSW_12_2_2"```
+- n-calls: int. The number of calls to be performed per evaluation for averaging. default: `100`
 
-- log-y: bool. Plot the y-axis values logarithmically; default: `False```.
+- input-type: str. Define the kind of inputs used during the measurement, either 'random', 'incremental', 'zeros', or a path to a root file. default: random.
 
-- bs-normalized: bool. Normalize the measured values with the batch size before plotting; default: `True```.
+- file_types: str. Comma-separated types of the output plot files. default: "pdf".
 
-- filling: bool. Plot the errors as error bands instead of error bars; default: `True```.
+- plot_postfix: str. An arbitrary postfix that is added with two underscores to all paths of produced plots. default: empty.
 
-- batch-sizes: int. The comma-separated list of batch sizes to be tested; default: `1,2,4```.
-
-- model-file: str. The absolute path of the json file containing the informations of the model to be tested.
-                   default: `$MLP_BASE/examples/model1/model.json```.
-
-- model-name: str. When set, use this name for the path used when storing outputs instead of a hashed version of
-        `--model-file```. default: empty.
-
-- input-files: str. comma-separated list of absolute paths of input files for the CMSSW analyzer (TODO: not implemented);
-                    when empty, random input values will be used; default: empty
-
-- events: int. The number of events to read from each input file for averaging measurements. default: `1```
-
-- repetitions: int. The number of repetitions to be performed per evaluation for averaging. default: `100```
-
-- warmup: int. The number of evaluations to be performed before starting the actual measurement.
-                    default: `10```
-
-- scram-arch: str. The SCRAM architecture used for the inference. default: `slc7_amd64_gcc10```
+- view_cmd: str. A command to execute after the task has run to visualize plots right in the terminal (e.g. "imgcat" if installed). default: empty.
 
 ## Output:
-- `runtime_plot__multiple_cmssw_{cmssw_version_1}_{cmssw_version_2}_{...}_different_batch_sizes_{batch_size_1}_{batch_size_2}_{...}.pdf```: The plot of the runtime measurement
-against the different batch sizes given.
+- `runtime_plot_networks_{param_1}_{param_2}_{...}_different_batch_sizes_{batch_size_1}_{batch_size_2}_{...}.pdf`: The plot of the runtime measurements against the different batch sizes given.
 
 ## Example:
 
 ```shell
-law run PlotRuntimesMultipleCMSSW --version test_multiple_cmssw \
-                                  --cmssw-versions CMSSW_12_2_4,CMSSW_12_2_2 \
-                                  --model-file $MLP_BASE/examples/model1/model.json \
-                                  --repetitions 500 \
-                                  --batch-sizes 1,2,4,8,16,32,64,128,256,512,1024 \
-                                  --log-y False
+law run PlotRuntimesMultipleParams --version test_several_networks \
+                                   --model-files $MLP_BASE/examples/simple_dnn/model.json,$MLP_BASE/examples/simple_dnn/model_onnx.json\
+                                   --model-names "dnn","dnn_onnx"\
+                                   --model-labels "dnn","dnn onnx"\
+                                   --cmssw-versions CMSSW_13_3_1,CMSSW_13_2_4 \
+                                   --n-calls 500 \
+                                   --batch-sizes 1,2,4,8,16,32,64,128,256,512,1024 \
+                                   --log-y False \
+                                   --bs-normalized True \
+                                   --top-right-label "$\sqrt{s}=13.6$ TeV"
 ```
 
 equivalent to the brace expanded version:
 
 ```shell
-law run PlotRuntimesMultipleCMSSW --version test_multiple_cmssw \
-                                  --cmssw-versions "CMSSW_12_2_{2,4}" \
-                                  --model-file $MLP_BASE/examples/model1/model.json \
-                                  --repetitions 500 \
-                                  --batch-sizes 1,2,4,8,16,32,64,128,256,512,1024 \
-                                  --log-y False
+law run PlotRuntimesMultipleParams --version test_several_networks \
+                                   --model-files "$MLP_BASE/examples/{simple_dnn/model,simple_dnn/model_onnx}.json"\
+                                   --model-names "dnn","dnn_onnx"\
+                                   --model-labels "dnn","dnn onnx"\
+                                   --cmssw-versions "CMSSW_13_{3_1,2_4}" \
+                                   --n-calls 500 \
+                                   --batch-sizes 1,2,4,8,16,32,64,128,256,512,1024 \
+                                   --log-y False \
+                                   --bs-normalized True \
+                                   --top-right-label "$\sqrt{s}=13.6$ TeV"
 ```
-
diff --git a/mlprof/tasks/runtime.py b/mlprof/tasks/runtime.py
index f1befa0..4658695 100644
--- a/mlprof/tasks/runtime.py
+++ b/mlprof/tasks/runtime.py
@@ -277,9 +277,8 @@ def __init__(self, *args, **kwargs):
 
         # create list of labels to plot
         self.params_product_params_to_write = [
-            combination_dict[key_to_write]
+            tuple([combination_dict[key_to_write] for key_to_write in self.params_to_write_labels])
             for combination_dict in self.labels_products_dict
-            for key_to_write in self.params_to_write_labels
         ]
 
     def flatten_tuple(self, value):