[Trainer] remove redundant memory metrics and set enable as default #8374
Conversation
|
Thanks for your contribution! |
paddlenlp/trainer/trainer.py
Outdated
| logs["current_memory_allocated"] = current_memory_allocated / divisor | ||
| logs["current_memory_reserved"] = current_memory_reserved / divisor | ||
| logs["max_memory_allocated"] = max_memory_allocated / divisor | ||
| logs["max_memory_reserved"] = max_memory_reserved / divisor |
There was a problem hiding this comment.
PaddleNLP/scripts/distribute/ci_case_dy.sh
Line 454 in 09a0ce7
这处也对应改了吧，gpu_max_memory_reserved -> max_memory_reserved
paddlenlp/trainer/trainer.py
Outdated
| max_memory_reserved = core.device_memory_stat_peak_value("Reserved", device_id) | ||
| logs["current_memory_allocated"] = current_memory_allocated / divisor | ||
| logs["current_memory_reserved"] = current_memory_reserved / divisor | ||
| logs["max_memory_allocated"] = max_memory_allocated / divisor |
There was a problem hiding this comment.
| logs["max_memory_allocated"] = max_memory_allocated / divisor | |
| logs["max_memory_allocated"] = max_memory_allocated >> 20 |
这个之前是MB单位，建议不要改变了，保持原来写法。用除法的话，是浮点数，还有小数位的问题，建议直接位运算，MB为单位
Codecov Report❌ Patch coverage is
❌ Your patch status has failed because the patch coverage (11.11%) is below the target coverage (80.00%). You can increase the patch coverage or adjust the target coverage. Additional details and impacted files@@ Coverage Diff @@
## develop #8374 +/- ##
===========================================
- Coverage 55.36% 54.60% -0.77%
===========================================
Files 614 636 +22
Lines 96016 108827 +12811
===========================================
+ Hits 53164 59428 +6264
- Misses 42852 49399 +6547 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
paddlenlp/trainer/training_args.py
Outdated
| ) | ||
| skip_memory_metrics: bool = field( | ||
| default=True, metadata={"help": "Whether or not to skip adding of memory profiler reports to metrics."} | ||
| default=False, metadata={"help": "Whether or not to skip adding of memory profiler reports to metrics."} |
|
This Pull Request is stale because it has been open for 60 days with no activity. 当前Pull Request 60天内无活动，被标记为stale。 |
|
This Pull Request is stale because it has been open for 60 days with no activity. 当前Pull Request 60天内无活动，被标记为stale。 |
|
|
|
This Pull Request is stale because it has been open for 60 days with no activity. 当前Pull Request 60天内无活动，被标记为stale。 |
|
Automatically closed by Paddle-bot. |
3 participants
PR types
Others
PR changes
Others
Description
remove redundant memory metrics and enable memory metrics print as default