-
Notifications
You must be signed in to change notification settings - Fork 210
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
Modifying the shard_num
read for templates to indicate that data_path
flag needs to be passed as an integer
#1037
Modifying the shard_num
read for templates to indicate that data_path
flag needs to be passed as an integer
#1037
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think extracting first integer from the string is a dangerous assumption. From my understanding, those data paths (e.g. data/c1
or data/collab2
) are not actual locations on disk (beyond the fact that user may also provide data/100shards/shard0
, wherein this regex will fail).
In this case, try casting data_path
to int
or raise a ValueError
that asks user to pass shard ID in the -d <shard_num>
format, since data path does not mean anything in this example.
|
shard_num
parameter to support default_data_path
valueshard_num
read for templates to indicate that data_path
flag needs to be passed as an integer
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Change the scope of try...except
block
@theakshaypant Please rebase your branch with |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
Signed-off-by: Pant, Akshay <[email protected]>
… shard num Signed-off-by: Pant, Akshay <[email protected]>
…ing it leads to an error when starting the collaborator for some templates Signed-off-by: Pant, Akshay <[email protected]>
…int) before loading dataset Signed-off-by: Pant, Akshay <[email protected]>
Signed-off-by: Pant, Akshay <[email protected]>
3035e0e
to
c6e35bf
Compare
…o akshay/fix-read_shard_from_default_path
Signed-off-by: Pant, Akshay <[email protected]>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM, with a couple of comments
$ fx collaborator create -n {COL_LABEL} -d {DATA_PATH:optional} | ||
$ fx collaborator create -n {COL_LABEL} -d {DATA_PATH} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
In general, the entire -d {DATA_PATH}
should be an optional parameter, to cover cases where the data path is already set in data.yaml
as part of the FL plan creation. Not sure what the :optional
refers to here though - is it just for the DATA_PATH? If so, then your suggestion is correct. If not, we should leave the documentation as is.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
In some examples (the ones modified by this PR), data path represents the shard number to be used for that collaborators. When we use the default value default_data_path = f"data/{collaborator_name}", an error is encountered when starting the collaborator is started.
Data = {'c1': 'data/c1', 'c2': 'data/c2'}
INFO 🧿 Starting a Collaborator Service. collaborator.py:99
INFO Building `src.dataloader.PyTorchMNISTInMemory` Module. plan.py:228
EXCEPTION : invalid literal for int() with base 10: 'data/c1'
Traceback (most recent call last):
File "/root/openfl-maestro/bin/fx", line 8, in <module>
sys.exit(entry())
File "/root/openfl-maestro/lib/python3.8/site-packages/openfl/interface/cli
.py", line 328, in entry
error_handler(e)
File "/root/openfl-maestro/lib/python3.8/site-packages/openfl/interface/cli
.py", line 246, in error_handler
raise error
File "/root/openfl-maestro/lib/python3.8/site-packages/openfl/interface/cli
.py", line 326, in entry
cli(max_content_width=120)
File "/root/openfl-maestro/lib/python3.8/site-packages/click/core.py", line
1157, in __call__
return self.main(*args, **kwargs)
File "/root/openfl-maestro/lib/python3.8/site-packages/click/core.py", line
1078, in main
rv = self.invoke(ctx)
File "/root/openfl-maestro/lib/python3.8/site-packages/click/core.py", line
1688, in invoke
return _process_result(sub_ctx.command.invoke(sub_ctx))
File "/root/openfl-maestro/lib/python3.8/site-packages/click/core.py", line
1688, in invoke
return _process_result(sub_ctx.command.invoke(sub_ctx))
File "/root/openfl-maestro/lib/python3.8/site-packages/click/core.py", line
1434, in invoke
return ctx.invoke(self.callback, **ctx.params)
File "/root/openfl-maestro/lib/python3.8/site-packages/click/core.py", line
783, in invoke
return __callback(*args, **kwargs)
File "/root/openfl-maestro/lib/python3.8/site-packages/openfl/interface/col
laborator.py", line 101, in start_
plan.get_collaborator(collaborator_name).run()
File "/root/openfl-maestro/lib/python3.8/site-packages/openfl/federated/pla
n/plan.py", line 601, in get_collaborator
data_loader = self.get_data_loader(collaborator_name)
File "/root/openfl-maestro/lib/python3.8/site-packages/openfl/federated/pla
n/plan.py", line 453, in get_data_loader
self.loader_ = Plan.build(**defaults)
File "/root/openfl-maestro/lib/python3.8/site-packages/openfl/federated/pla
n/plan.py", line 235, in build
instance = getattr(module, class_name)(**settings)
File "/home/akshay/akshay/src/dataloader.py", line 30, in __init__
shard_num=int(data_path), **kwargs
ValueError: invalid literal for int() with base 10: 'data/c1'
Hence at the time of creating this PR, I suggested modifying the documentation to remove the "optional" for DATA_PATH.
We can keep the documentation as is with a more descriptive error message in the example as you suggested in another comment.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Reverted change in aae8185.
try: | ||
int(data_path) | ||
except: | ||
raise ValueError("Expected `%s` to be representable as `int`.", data_path) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
In the error message, specify that this is for the shard number for the given collaborator.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Addressed in ca90679.
try: | ||
int(data_path) | ||
except: | ||
raise ValueError("Expected `%s` to be representable as `int`.", data_path) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Same comment here, and for the other ValueError
instances throughout the PR
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Addressed in ca90679.
…o akshay/fix-read_shard_from_default_path
Signed-off-by: Pant, Akshay <[email protected]>
Signed-off-by: Pant, Akshay <[email protected]>
Description
This is a documentation patch.
Currently, docs seem to show
data_path
to be an optional flag. But for the specific example, it is mandatory. This PR makes the mention explicit.In addition, a guardrail to expect
data_path
as a shard number is added in all affected tutorials.PR Changes