-
Notifications
You must be signed in to change notification settings - Fork 74
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add option for mixed precision training #121
Comments
edward-bot
pushed a commit
that referenced
this issue
Dec 8, 2019
Code was not actually using mixed precision. It's now settable via a flag. Fixes #121. Results are comparable with and without. Speedup is insignificant however, so I'm leaving it off by default. PiperOrigin-RevId: 284398786
edward-bot
pushed a commit
that referenced
this issue
Dec 8, 2019
Code was not actually using mixed precision. It's now settable via a flag. Fixes #121. Results are comparable with and without. Speedup is insignificant however, so I'm leaving it off by default. PiperOrigin-RevId: 284398786
edward-bot
pushed a commit
that referenced
this issue
Dec 10, 2019
Code was not actually using mixed precision. It's now settable via a flag. Fixes #121. Results are comparable with and without. Speedup is insignificant however, so I'm leaving it off by default. PiperOrigin-RevId: 284398786
edward-bot
pushed a commit
that referenced
this issue
Dec 10, 2019
Code was not actually using mixed precision. It's now settable via a flag. Fixes #121. Results are comparable with and without. Speedup is insignificant however, so I'm leaving it off by default. PiperOrigin-RevId: 284398786
edward-bot
pushed a commit
that referenced
this issue
Dec 10, 2019
Code was not actually using mixed precision. It's now settable via a flag. Fixes #121. Results are comparable with and without. Speedup is insignificant however, so I'm leaving it off by default. PiperOrigin-RevId: 284398786
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Deterministic and BatchEnsemble baselines cast data as bfloat16 by default with TPUs. Following the cloud TPU imagenet example, we need to set a policy if we want to maintain bfloat16 for the activations, etc.
It should be easy enough to set up a boolean flag to use bfloat16 with that policy or otherwise operate in float32.
The text was updated successfully, but these errors were encountered: