train.md

# Train

In this tutorial, we will introduce how to use the scripts provided in MMPretrain to start a training task. If
you need, we also have some practice examples about [how to pretrain with custom dataset](../notes/pretrain_custom_dataset.md)
and [how to finetune with custom dataset](../notes/finetune_custom_dataset.md).

## Train with your PC

You can use `tools/train.py` to train a model on a single machine with a CPU and optionally a GPU.

Here is the full usage of the script:

```shell
python tools/train.py ${CONFIG_FILE} [ARGS]
```

````{note}
By default, MMPretrain prefers GPU to CPU. If you want to train a model on CPU, please empty `CUDA_VISIBLE_DEVICES` or set it to -1 to make GPU invisible to the program.

```bash
CUDA_VISIBLE_DEVICES=-1 python tools/train.py ${CONFIG_FILE} [ARGS]
```
````

| ARGS                                  | Description                                                                                                                                                         |
| ------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `CONFIG_FILE`                         | The path to the config file.                                                                                                                                        |
| `--work-dir WORK_DIR`                 | The target folder to save logs and checkpoints. Defaults to a folder with the same name of the config file under `./work_dirs`.                                     |
| `--resume [RESUME]`                   | Resume training. If specify a path, resume from it, while if not specify, try to auto resume from the latest checkpoint.                                            |
| `--amp`                               | Enable automatic-mixed-precision training.                                                                                                                          |
| `--no-validate`                       | **Not suggested**. Disable checkpoint evaluation during training.                                                                                                   |
| `--auto-scale-lr`                     | Auto scale the learning rate according to the actual batch size and the original batch size.                                                                        |
| `--no-pin-memory`                     | Whether to disable the `pin_memory` option in dataloaders.                                                                                                          |
| `--no-persistent-workers`             | Whether to disable the `persistent_workers` option in dataloaders.                                                                                                  |
| `--cfg-options CFG_OPTIONS`           | Override some settings in the used config, the key-value pair in xxx=yyy format will be merged into the config file. If the value to be overwritten is a list, it should be of the form of either `key="[a,b]"` or `key=a,b`. The argument also allows nested list/tuple values, e.g. `key="[(a,b),(c,d)]"`. Note that the quotation marks are necessary and that no white space is allowed. |
| `--launcher {none,pytorch,slurm,mpi}` | Options for job launcher.                                                                                                                                           |

## Train with multiple GPUs

We provide a shell script to start a multi-GPUs task with `torch.distributed.launch`.

```shell
bash ./tools/dist_train.sh ${CONFIG_FILE} ${GPU_NUM} [PY_ARGS]
```

| ARGS          | Description                                                                        |
| ------------- | ---------------------------------------------------------------------------------- |
| `CONFIG_FILE` | The path to the config file.                                                       |
| `GPU_NUM`     | The number of GPUs to be used.                                                     |
| `[PY_ARGS]`   | The other optional arguments of `tools/train.py`, see [here](#train-with-your-pc). |

You can also specify extra arguments of the launcher by environment variables. For example, change the
communication port of the launcher to 29666 by the below command:

```shell
PORT=29666 bash ./tools/dist_train.sh ${CONFIG_FILE} ${GPU_NUM} [PY_ARGS]
```

If you want to startup multiple training jobs and use different GPUs, you can launch them by specifying
different ports and visible devices.

```shell
CUDA_VISIBLE_DEVICES=0,1,2,3 PORT=29500 bash ./tools/dist_train.sh ${CONFIG_FILE1} 4 [PY_ARGS]
CUDA_VISIBLE_DEVICES=4,5,6,7 PORT=29501 bash ./tools/dist_train.sh ${CONFIG_FILE2} 4 [PY_ARGS]
```

## Train with multiple machines

### Multiple machines in the same network

If you launch a training job with multiple machines connected with ethernet, you can run the following commands:

On the first machine:

```shell
NNODES=2 NODE_RANK=0 PORT=$MASTER_PORT MASTER_ADDR=$MASTER_ADDR bash tools/dist_train.sh $CONFIG $GPUS
```

On the second machine:

```shell
NNODES=2 NODE_RANK=1 PORT=$MASTER_PORT MASTER_ADDR=$MASTER_ADDR bash tools/dist_train.sh $CONFIG $GPUS
```

Comparing with multi-GPUs in a single machine, you need to specify some extra environment variables:

| ENV_VARS      | Description                                                                  |
| ------------- | ---------------------------------------------------------------------------- |
| `NNODES`      | The total number of machines.                                                |
| `NODE_RANK`   | The index of the local machine.                                              |
| `PORT`        | The communication port, it should be the same in all machines.               |
| `MASTER_ADDR` | The IP address of the master machine, it should be the same in all machines. |

Usually it is slow if you do not have high speed networking like InfiniBand.

### Multiple machines managed with slurm

If you run MMPretrain on a cluster managed with [slurm](https://slurm.schedmd.com/), you can use the script `tools/slurm_train.sh`.

```shell
[ENV_VARS] ./tools/slurm_train.sh ${PARTITION} ${JOB_NAME} ${CONFIG_FILE} ${WORK_DIR} [PY_ARGS]
```

Here are the arguments description of the script.

| ARGS          | Description                                                                        |
| ------------- | ---------------------------------------------------------------------------------- |
| `PARTITION`   | The partition to use in your cluster.                                              |
| `JOB_NAME`    | The name of your job, you can name it as you like.                                 |
| `CONFIG_FILE` | The path to the config file.                                                       |
| `WORK_DIR`    | The target folder to save logs and checkpoints.                                    |
| `[PY_ARGS]`   | The other optional arguments of `tools/train.py`, see [here](#train-with-your-pc). |

Here are the environment variables can be used to configure the slurm job.

| ENV_VARS        | Description                                                                                                |
| --------------- | ---------------------------------------------------------------------------------------------------------- |
| `GPUS`          | The number of GPUs to be used. Defaults to 8.                                                              |
| `GPUS_PER_NODE` | The number of GPUs to be allocated per node..                                                              |
| `CPUS_PER_TASK` | The number of CPUs to be allocated per task (Usually one GPU corresponds to one task). Defaults to 5.      |
| `SRUN_ARGS`     | The other arguments of `srun`. Available options can be found [here](https://slurm.schedmd.com/srun.html). |