Update docs (#2118)

* Update mlcflow commands

Author: arjunsuresh

.github/workflows/auto-update-dev.yml

@@ -7,6 +7,7 @@ on:
 
 jobs:
   update-dev:
+    if: github.repository_owner == 'mlcommons'
     strategy:
       matrix: 
         branch: [ "dev", "docs" ]

.github/workflows/build_wheels.yml

@@ -70,6 +70,7 @@ jobs:
   
   build_wheels:
     name: Build wheels on ${{ matrix.os }}
+    if: github.repository_owner == 'mlcommons'
     needs: update_version
     runs-on: ${{ matrix.os }}
     strategy:
@@ -101,6 +102,7 @@ jobs:
           path: wheels
 
   publish_wheels:
+    if: github.repository_owner == 'mlcommons'
     needs: build_wheels  # Wait for the build_wheels job to complete
     runs-on: ubuntu-latest  # Only run this job on Linux
     environment: release

.github/workflows/publish.yaml

@@ -9,6 +9,7 @@ on:
     branches:
       - master
       - docs
+      - dev
 
 jobs:
 

docs/benchmarks/automotive/3d_object_detection/get-pointpainting-data.md

@@ -3,7 +3,27 @@ hide:
   - toc
 ---
 
-# 3D Object Detection using PointPainting
+# 3-D Object Detection using PointPainting
 
-TBD
+## Dataset
 
+> **Note:** By default, the waymo dataset is downloaded from the mlcommons official drive. One has to accept the [MLCommons Waymo Open Dataset EULA](https://waymo.mlcommons.org/) to access the dataset files. 
+
+The benchmark implementation run command will automatically download the preprocessed dataset. In case you want to download only the datasets, you can use the below command.
+
+```bash
+mlcr get,dataset,waymo -j
+```
+
+- `--outdirname=<PATH_TO_DOWNLOAD_WAYMO_DATASET>` could be provided to download the dataset to a specific location.
+
+## Model
+> **Note:** By default, the PointPainting is downloaded from the mlcommons official drive. One has to accept the [MLCommons Waymo Open Dataset EULA](https://waymo.mlcommons.org/) to access the model files. 
+
+The benchmark implementation run command will automatically download the model. In case you want to download only the PointPainting model, you can use the below command.
+
+```bash
+mlcr get,ml-model,pointpainting -j
+```
+
+- `--outdirname=<PATH_TO_DOWNLOAD_POINTPAINTING_MODEL>` could be provided to download the model files to a specific location.

docs/benchmarks/language/get-llama2-70b-data.md

@@ -26,10 +26,10 @@ Get the Official MLPerf LLAMA2-70b Model
 
     ### Pytorch
     ```
-    mlcr get,ml-model,llama2-70b,_pytorch -j
+    mlcr get,ml-model,llama2-70b,_pytorch -j --outdirname=<My download path>
     ```
 
 !!! tip
 
-    Downloading llama2-70B model from Hugging Face will prompt you to enter the Hugging Face username and password. Please note that the password required is the [**access token**](https://huggingface.co/settings/tokens) generated for your account. Additionally, ensure that your account has access to the [llama2-70B](https://huggingface.co/meta-llama/Llama-2-70b-chat-hf) model.
+    [Access Request Link](https://llama2.mlcommons.org/) for MLCommons members
 

docs/benchmarks/language/get-llama3_1-405b-data.md

@@ -32,10 +32,9 @@ Get the Official MLPerf LLAMA3.1-405b Model
 
     ### Pytorch
     ```
-    mlcr get,ml-model,llama3 --outdirname=<path to download> --hf_token=<huggingface access token> -j
+    mlcr get,ml-model,llama3 --outdirname=<path to download>  -j
     ```
 
 !!! tip
 
-    Downloading llama3.1-405B model from Hugging Face will require an [**access token**](https://huggingface.co/settings/tokens) which could be generated for your account. Additionally, ensure that your account has access to the [llama3.1-405B](https://huggingface.co/meta-llama/Llama-3.1-405B-Instruct) model. 
-
+    [Access Request Link](https://llama3-1.mlcommons.org/) for MLCommons members

docs/changelog/changelog.md

@@ -1,2 +1,19 @@
-# Release Notes
+# **Release Notes**  
 
+🚀 **mlc-scripts 1.0.0** was released on **February 10, 2025**, introducing full support for **MLPerf Inference v5.0** using the first stable release of **MLCFlow 1.0.1**.  
+
+🔹 All previous **CM scripts** used in MLPerf have been successfully **ported to the MLC interface**, ensuring seamless integration. Additionally, all **GitHub Actions** are now passing, confirming a stable and reliable implementation.  
+
+## **Key Updates in MLCFlow**  
+
+✅ **Simplified Interface**  
+- A redesigned approach using **Actions and Targets**, making the CLI more intuitive for users.  
+
+✅ **Unified Automation Model**  
+- Consolidated into a **single automation entity**: **Script**, which is seamlessly extended by **Cache, Docker, and Tests**.  
+
+✅ **Improved Docker Integration**  
+- A **cleaner, more efficient Docker extension**, streamlining containerized execution.  
+
+✅ **Enhanced Script Management**  
+- Tighter integration between the interface and script automation, making script **creation and management easier than ever**.  

docs/changelog/index.md

@@ -3,5 +3,10 @@ hide:
   - toc
 ---
 
-# What's New, What's Coming
+# What's New & What's Coming 🚀  
 
+!!! info  
+    **Inference v5.0 Submission** is approaching! The submission deadline is **February 28, 2025, at 1 PM PST**.  
+
+!!! tip  
+    Starting **January 2025**, MLPerf Inference automations are powered by **[MLCFlow](https://docs.mlcommons.org/mlcflow)**—a newly developed Python package that replaces the previously used **CMind** package. This transition enhances automation, streamlines workflows, and makes MLPerf scripts more independent and simpler.  

docs/power/index.md

@@ -0,0 +1,93 @@
+# Power Measurement
+
+*Originally Prepared by the MLCommons taskforce on automation and reproducibility and [OctoML](https://octoml.ai)*.
+
+## Requirements
+
+1. Power analyzer (anyone [certified by SPEC PTDaemon](https://www.spec.org/power/docs/SPECpower-Device_List.html)). Yokogawa is the one that most submitters have submitted with and a new single-channel model like 310E can cost around 3000$. 
+
+2. SPEC PTDaemon (can be downloaded from [here](https://github.com/mlcommons/power) after signing the EULA which can be requested by sending an email to `[email protected]`). Once you have GitHub access to the MLCommons power repository then the MLC workflow will automatically download and configure the SPEC PTDaemon tool.
+
+3. Access to the [MLCommons power-dev](https://github.com/mlcommons/power-dev) repository which has the `server.py` to be run on the director node and `client.py` to be run on the SUT node. This repository being public will be automatically pulled by the MLC workflow.
+
+## Connecting power analyzer to the computer
+
+We need to connect the power analyzer to a director machine via USB if the director machine is running Linux. Ethernet and serial modes are supported only on Windows. The power supply to the SUT is done through the power analyzer (current in series and voltage in parallel). An adapter like [this](https://amzn.to/3Cl2TV5) can help avoid cutting the electrical wires. 
+
+![pages (14)](https://user-images.githubusercontent.com/4791823/210117283-82375460-5b3a-4e8a-bd85-9d33675a5843.png).
+
+The director machine runs the `server.py` script and loads a server process that communicates with the SPEC PTDaemon. When a client connects to it (using `client.py`), it in turn connects to the PTDaemon and initiates a measurement run. Once the measurement ends, the power log files are transferred to the client. 
+
+## Ranging mode and Testing mode
+
+Power analyzers usually have different current and voltage ranges it supports and the exact ranges to be used depends on a given SUT and this needs some empirical data. We can do a ranging run where the current and voltage ranges are set to `Auto` and the power analyzer automatically figures out the correct ranges needed. These determined ranges are then used for a proper testing mode run. Using the 'auto' mode in a testing run is not allowed as it can mess up the measurements.
+
+
+## Start Power Server (Power analyzer should be connected to this computer and PTDaemon runs here)
+
+If you are having GitHub access to [MLCommons power](https://github.com/mlcommons/power-dev) repository, PTDaemon should be automatically installed using the below MLC command:
+
+PS: The below command will ask for `sudo` permission on Linux and should be run with administrator privilege on Windows (to do NTP time sync).
+```bash
+mlcr mlperf,power,server --device_type=49 --device_port=/dev/usbtmc0
+```
+* ``--interface_flag="-U" and `--device_port=1` (can change as per the USB slot used for connecting) can be used on Windows for USB connection
+* `--device_type=49` corresponds to Yokogawa 310E and `ptd -h` should list the device_type for all supported devices. The location of `ptd` can be found using the below command
+* `--device_port=20` and `--interface_flag="-g" can be used to connect to GPIB interface (currently supported only on Windows) with the serial address set to 20
+```bash
+cat `mlc find cache --tags=get,spec,ptdaemon`/mlc-cached-state.json
+```
+
+An example analyzer configuration file
+```
+[server]
+ntpserver = time.google.com
+listen = 0.0.0.0 4950
+
+[ptd]
+ptd = C:\Users\arjun\CM\repos\local\cache\5a0a52d578724774\repo\PTD\binaries\ptd-windows-x86.exe
+analyzerCount = 2
+[analyzer2]
+interfaceflag = -g
+devicetype = 8
+deviceport = 20
+networkport = 8888
+
+[analyzer1]
+interfaceflag = -y
+devicetype = 49
+deviceport = C3YD21068E
+networkport = 8889
+```
+
+### Running the power server inside a docker container
+
+```bash
+mlc docker script --tags=run,mlperf,power,server --docker_gh_token=<GITHUB AUTH_TOKEN> \
+--device=/dev/usbtmc0
+```
+* Device address may need to be changed depending on the USB port being used
+* The above command uses a host-container port mapping 4950:4950 which can be changed by using `--docker_port_maps,=4950:4950`
+
+## Running a dummy workload with power (on host machine)
+
+```bash
+mlcr mlperf,power,client --power_server=<POWER_SERVER_IP> 
+```
+
+### Run a dummy workload with power inside a docker container
+
+```bash
+mlc docker script --tags==mlperf,power,client --power_server=<POWER_SERVER_IP>"
+```
+
+## Running MLPerf Image Classification with power
+
+```bash
+mlcr app,mlperf,inference,_reference,_power,_resnet50,_onnxruntime,_cpu --mode=performance --power_server=<POWER_SERVER_IP>
+```
+
+### Running MLPerf Image Classification with power inside a docker container
+```bash
+mlcr app,mlperf,inference,_reference,_power,_resnet50,_onnxruntime,_cpu --mode=performance --power_server=<POWER_SERVER_IP> --docker
+```

docs/submission/index.md

@@ -73,7 +73,7 @@ flowchart LR
       direction TB
       A[populate system details] --> B[generate submission structure]
       B --> C[truncate-accuracy-logs]
-      C --> D{Infer low talency results <br>and/or<br> filter out invalid results}
+      C --> D{Infer low latency results <br>and/or<br> filter out invalid results}
       D --> yes --> E[preprocess-mlperf-inference-submission]
       D --> no --> F[run-mlperf-inference-submission-checker]
       E --> F
@@ -184,7 +184,7 @@ Once you have all the results on the system, you can upload them to the MLCommon
 === "via CLI"
     You can do the following command which will run the submission checker and upload the results to the MLCommons submission server
     ```
-    mlcr run,submission,checker \
+    mlcr run,submission,checker,inference \
     --submitter_id=<> \
     --submission_dir=<Path to the submission folder>
     ```