Config for TPU pod #239
Open
Config for TPU pod #239
+2
−1
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
I ran `train.py` on a TPU pod v3-256 and got the following error:
ValueError: TPUConfig.num_shards is not set correctly ....
Found in https://cloud.google.com/tpu/docs/training-on-tpu-pods#providing_the_tpu_name_and_region_to_tpuclusterresolver that
> For single device training, you can specify either the TPU name or an IP address, for example: `grpc://1.2.3.4:8470`.
> For TPU Pods you must use the TPU name so that TensorFlow can discover the IP addresses of all the hosts available for training distribution.
So, in the case of a TPU pod, setting `master` doesn't work. I just tried setting `cluster` and it worked, all 32 hosts in the TPU pod were detected and used correctly.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
I ran
train.pyon a TPU pod v3-256 and got the following error:Found in https://cloud.google.com/tpu/docs/training-on-tpu-pods#providing_the_tpu_name_and_region_to_tpuclusterresolver that
So, in the case of a TPU pod, setting
masterdoesn't work. I just tried settingclusterand it worked, all 32 hosts in the TPU pod were detected and used correctly.