Evaluating and comparing the performance from different setups can be quite tricky if you don’t know what to look for. For example, you cannot run the same script with the same batch size across TPU, multi-GPU, and single-GPU with Accelerate and expect your results to line up.
There’s three reasons for this that this tutorial will cover:
- Setting the right seeds
- Observed Batch Sizes
- Learning Rates
While this issue has not come up as much, make sure to use utils.set_seed() to fully set the seed in all distributed cases so training will be reproducable:
from accelerate.utils import set_seed set_seed(42)
Why is this important? Under the hood this will set 5 different seed settings:
random.seed(seed) np.random.seed(seed) torch.manual_seed(seed) torch.cuda.manual_seed_all(seed) # ^^ safe to call this function even if cuda is not available if is_tpu_available(): xm.set_rng_state(seed)
The random state, numpy’s state, torch, torch’s cuda state, and if TPUs are available torch_xla’s cuda state.
When training with Accelerate, the batch size passed to the dataloader is the batch size per GPU. What this entails is a batch size of 64 on two GPUs is truly a batch size of 128. As a result, when testing on a single GPU this needs to be accounted for, as well as similarly for TPUs.
The below table can be used as a quick reference to try out different batch sizes:
In this example there are two GPUs for “Multi-GPU” and a TPU pod with 8 workers
|Single GPU Batch Size||Multi-GPU Equivalent Batch Size||TPU Equivalent Batch Size|
Since users can have their own learning rate schedulers defined, we leave this up to the user to decide if they wish to scale their learning rate or not.
learning_rate = 1e-3 accelerator = Accelerator() learning_rate *= accelerator.num_processes optimizer = AdamW(params=model.parameters(), lr=learning_rate)
You will also find that
accelerate will step the learning rate based on the number of processes being trained on. This is because
of the observed batch size noted earlier. So in a case of 2 GPUs, the learning rate will be stepped twice as often as a single GPU
to account for the batch size being twice as large (if no changes to the batch size on the single GPU instance are made).