Skip to content

Conversation

ultrons
Copy link

@ultrons ultrons commented Nov 7, 2023

Added steptime and model flop estimation calc.

data, fsdp, mdl = self.args.spmd_mesh.ici_mesh_shape
num_devices = data * fsdp * mdl
num_tokens = inputs["input_ids"].numel() / num_devices
xm.master_print(f"Step time: {step_time}: Model TFLOPS: {self.model_flops(step_time, num_tokens)}")
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

cc @yeounoh, we can emit this to Tensorboard for perf tests (in addition to the convergence metrics)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants