-
Notifications
You must be signed in to change notification settings - Fork 1.2k
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
exp init: add simple dvc exp init command #6621
Changes from all commits
File filter
Filter by extension
Conversations
Jump to
Diff view
Diff view
There are no files selected for viewing
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -1,5 +1,6 @@ | ||
import argparse | ||
import logging | ||
import os | ||
from collections import Counter, OrderedDict, defaultdict | ||
from datetime import date, datetime | ||
from fnmatch import fnmatch | ||
|
@@ -789,6 +790,62 @@ def run(self): | |
return 0 | ||
|
||
|
||
class CmdExperimentsInit(CmdBase): | ||
CODE = "src" | ||
DATA = "data" | ||
MODELS = "models" | ||
DEFAULT_METRICS = "metrics.json" | ||
DEFAULT_PARAMS = "params.yaml" | ||
PLOTS = "plots" | ||
DVCLIVE = "dvclive" | ||
DEFAULT_NAME = "default" | ||
|
||
def run(self): | ||
from dvc.command.stage import parse_cmd | ||
|
||
cmd = parse_cmd(self.args.cmd) | ||
if not cmd: | ||
raise InvalidArgumentError("command is not specified") | ||
if self.args.interactive: | ||
raise NotImplementedError( | ||
"'-i/--interactive' is not implemented yet." | ||
) | ||
if self.args.explicit: | ||
raise NotImplementedError("'--explicit' is not implemented yet.") | ||
if self.args.template: | ||
raise NotImplementedError("template is not supported yet.") | ||
|
||
from dvc.utils.serialize import LOADERS | ||
|
||
code = self.args.code or self.CODE | ||
data = self.args.data or self.DATA | ||
models = self.args.models or self.MODELS | ||
metrics = self.args.metrics or self.DEFAULT_METRICS | ||
params_path = self.args.params or self.DEFAULT_PARAMS | ||
plots = self.args.plots or self.PLOTS | ||
dvclive = self.args.live or self.DVCLIVE | ||
|
||
_, ext = os.path.splitext(params_path) | ||
params = list(LOADERS[ext](params_path)) | ||
|
||
name = self.args.name or self.DEFAULT_NAME | ||
stage = self.repo.stage.add( | ||
name=name, | ||
cmd=cmd, | ||
deps=[code, data], | ||
outs=[models], | ||
params=[{params_path: params}], | ||
metrics_no_cache=[metrics], | ||
plots_no_cache=[plots], | ||
live=dvclive, | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. Default stage can ignore There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. We don't have a default stage/template yet, so I left it with the default. I'll change this when we have templates. |
||
force=True, | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. I think Edit: It might be unclear to a new user that this may overwrite an existing stage. There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. Also a note that we obviously need to revisit this later to generalize for other templates. There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. I'll add the |
||
) | ||
|
||
if self.args.run: | ||
return self.repo.experiments.run(targets=[stage.addressing]) | ||
return 0 | ||
|
||
|
||
def add_parser(subparsers, parent_parser): | ||
EXPERIMENTS_HELP = "Commands to run and compare experiments." | ||
|
||
|
@@ -1303,6 +1360,68 @@ def add_parser(subparsers, parent_parser): | |
) | ||
experiments_remove_parser.set_defaults(func=CmdExperimentsRemove) | ||
|
||
EXPERIMENTS_INIT_HELP = "Initialize experiments." | ||
experiments_init_parser = experiments_subparsers.add_parser( | ||
"init", | ||
parents=[parent_parser], | ||
description=append_doc_link(EXPERIMENTS_INIT_HELP, "exp/init"), | ||
formatter_class=argparse.RawDescriptionHelpFormatter, | ||
) | ||
experiments_init_parser.add_argument( | ||
"cmd", | ||
nargs=argparse.REMAINDER, | ||
help="Command to execute.", | ||
metavar="command", | ||
) | ||
experiments_init_parser.add_argument( | ||
"--run", | ||
action="store_true", | ||
help="Run the experiment after initializing it", | ||
) | ||
experiments_init_parser.add_argument( | ||
"--interactive", | ||
"-i", | ||
action="store_true", | ||
help="Prompt for values that are not provided", | ||
) | ||
experiments_init_parser.add_argument( | ||
"--template", help="Stage template to use to fill with provided values" | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. I think we should include the default value in the help message (and for all non-boolean options below). There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. Will add defaults in the upcoming PRs, as the place where I'll keep those defaults may constantly change. :) |
||
) | ||
experiments_init_parser.add_argument( | ||
"--explicit", help="Only use the path values explicitly provided" | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. By the way, feel free to suggest a better name for this option. Maybe |
||
) | ||
experiments_init_parser.add_argument( | ||
"--name", "-n", help="Name of the stage to create" | ||
) | ||
experiments_init_parser.add_argument( | ||
"--code", | ||
help="Path to the source file or directory " | ||
"which your experiments depend", | ||
) | ||
experiments_init_parser.add_argument( | ||
"--data", | ||
help="Path to the data file or directory " | ||
"which your experiments depend", | ||
) | ||
experiments_init_parser.add_argument( | ||
"--models", | ||
help="Path to the model file or directory for your experiments", | ||
) | ||
experiments_init_parser.add_argument( | ||
"--params", help="Path to the parameters file for your experiments" | ||
) | ||
experiments_init_parser.add_argument( | ||
"--metrics", help="Path to the metrics file for your experiments" | ||
) | ||
experiments_init_parser.add_argument( | ||
"--plots", | ||
help="Path to the plots file or directory for your experiments", | ||
) | ||
experiments_init_parser.add_argument( | ||
"--live", help="Path to log dvclive outputs for your experiments" | ||
) | ||
experiments_init_parser.set_defaults(func=CmdExperimentsInit) | ||
|
||
|
||
def _add_run_common(parser): | ||
"""Add common args for 'exp run' and 'exp resume'.""" | ||
|
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,34 @@ | ||
import os | ||
|
||
from dvc.command.experiments import CmdExperimentsInit | ||
from dvc.main import main | ||
from dvc.utils.serialize import load_yaml | ||
|
||
|
||
def test_init(tmp_dir, dvc): | ||
tmp_dir.gen( | ||
{ | ||
CmdExperimentsInit.CODE: {"copy.py": ""}, | ||
"data": "data", | ||
"params.yaml": '{"foo": 1}', | ||
"dvclive": {}, | ||
"plots": {}, | ||
} | ||
) | ||
code_path = os.path.join(CmdExperimentsInit.CODE, "copy.py") | ||
script = f"python {code_path}" | ||
|
||
assert main(["exp", "init", script]) == 0 | ||
assert load_yaml(tmp_dir / "dvc.yaml") == { | ||
"stages": { | ||
"default": { | ||
"cmd": script, | ||
"deps": ["data", "src"], | ||
"live": {"dvclive": {"html": True, "summary": True}}, | ||
"metrics": [{"metrics.json": {"cache": False}}], | ||
"outs": ["models"], | ||
"params": ["foo"], | ||
"plots": [{"plots": {"cache": False}}], | ||
} | ||
} | ||
} |
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.
It doesn't need to block this PR, but see #6605 for proposal to eliminate need for parameter parsing here.