add memory metrics to TensorBoard #60
Conversation
[ghstack-poisoned]
facebook-github-bot
added
the
CLA Signed
train.py
Outdated
| "global_avg_loss": global_avg_loss, | ||
| "global_max_loss": global_max_loss, | ||
| "loss/global_avg": global_avg_loss, | ||
| "loss/global_max": global_max_loss, |
There was a problem hiding this comment.
nit - using the / here is confusing to me...I thought it represented the loss divided by the global avg, and same for max...
maybe consider just an _ or : or even :: as the separator? (loss:global_avg, loss::global_max, memory_current_active).
There was a problem hiding this comment.
oh good point. I use [tag]/[metric] here because TB collects plots under the same [tag] together in a row, so that they form a visual group. Just like in the picture in PR summary, memory metrics are grouped into memory_current, and memory_peak. I'll explore a way that can achieve this but without ambiguity for losses.
There was a problem hiding this comment.
Did some exploration, e.g. tried to put related metrics into a single plot. The options we have are add_scalars and add_custom_scalars, and it seems neither is ideal (e.g.). I'm changing loss/global_avg to loss_metrics/global_avg for now to make it less ambiguous.
lessw2020
left a comment
lessw2020
left a comment
There was a problem hiding this comment.
looks great, thanks for integrating these stats!
one very minor nit about the / being possibly confused as division when used in labelling.
<img width="1391" alt="Screenshot 2024-02-15 at 5 19 09 PM" src="https://github.com/pytorch-labs/torchtrain/assets/150487191/af8a2efb-13ff-4e8f-84f2-b245784747ed"> [ghstack-poisoned]
ghstack-source-id: da7e02b Pull Request resolved: pytorch#60
Stack from ghstack (oldest at bottom):