Track API usage #2706
Track API usage #2706
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/2706
Note: Links to docs will display an error until the docs builds have been completed. ⏳ No Failures, 3 PendingAs of commit 80ee54a with merge base a1a9632 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
bot
added
the
CLA Signed
andrewor14
force-pushed
the
track-api-usage
branch
from
ccd5512 to
f6847de
Compare
andrewor14
added
the
topic: improvement
| @@ -101,6 +102,7 @@ def convert_to_float8_training( | |||
| Returns: | |||
| nn.Module: The modified module with swapped linear layers. | |||
| """ | |||
| torch._C._log_api_usage_once("torchao.quantization.convert_to_float8_training") | |||
There was a problem hiding this comment.
@vkuzo I put this under the torchao.quantization namespace based on some past discussions on code reorg that hasn't happened yet. Do you prefer this namespace compared to torchao.float8, which is the current location?
There was a problem hiding this comment.
also any other float8 APIs you'd like to track?
There was a problem hiding this comment.
is the reason for preemptively logging here ahead of the eventual migration so that querying usage metrics is easier (i.e., no need to aggregate results of 2 queries)?
There was a problem hiding this comment.
yeah and also we can start tracking earlier, not sure when the migration will happen
There was a problem hiding this comment.
IMO better to match existing code structure, and if a refactor happens we'll deal with the logging change. I say this because there is no timeline or owner for actually migrating torchao.float8 at the moment.
yeah I think so |
andrewor14
force-pushed
the
track-api-usage
branch
from
f6847de to
e700c81
Compare
No description provided.