hydraflow 0.2.2__py3-none-any.whl → 0.2.4__py3-none-any.whl
Sign up to get free protection for your applications and to get access to all the features.
- hydraflow/__init__.py +7 -13
- hydraflow/asyncio.py +199 -0
- hydraflow/config.py +3 -3
- hydraflow/context.py +72 -21
- hydraflow/mlflow.py +64 -12
- hydraflow/runs.py +428 -190
- {hydraflow-0.2.2.dist-info → hydraflow-0.2.4.dist-info}/METADATA +5 -6
- hydraflow-0.2.4.dist-info/RECORD +10 -0
- hydraflow-0.2.2.dist-info/RECORD +0 -9
- {hydraflow-0.2.2.dist-info → hydraflow-0.2.4.dist-info}/WHEEL +0 -0
- {hydraflow-0.2.2.dist-info → hydraflow-0.2.4.dist-info}/licenses/LICENSE +0 -0
hydraflow/runs.py
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
"""
|
2
|
-
This module provides functionality for managing and interacting with MLflow
|
3
|
-
It includes the `RunCollection` class and various methods to filter
|
4
|
-
retrieve run information, log artifacts, and load configurations.
|
2
|
+
This module provides functionality for managing and interacting with MLflow
|
3
|
+
runs. It includes the `RunCollection` class and various methods to filter
|
4
|
+
runs, retrieve run information, log artifacts, and load configurations.
|
5
5
|
"""
|
6
6
|
|
7
7
|
from __future__ import annotations
|
@@ -37,34 +37,49 @@ def search_runs(
|
|
37
37
|
"""
|
38
38
|
Search for Runs that fit the specified criteria.
|
39
39
|
|
40
|
-
This function wraps the `mlflow.search_runs` function and returns the
|
41
|
-
as a `RunCollection` object. It allows for flexible searching of
|
42
|
-
various criteria.
|
40
|
+
This function wraps the `mlflow.search_runs` function and returns the
|
41
|
+
results as a `RunCollection` object. It allows for flexible searching of
|
42
|
+
MLflow runs based on various criteria.
|
43
43
|
|
44
44
|
Note:
|
45
45
|
The returned runs are sorted by their start time in ascending order.
|
46
46
|
|
47
47
|
Args:
|
48
|
-
experiment_ids: List of experiment IDs. Search can
|
49
|
-
experiment names, but not both in the
|
50
|
-
``None`` or ``[]`` will result in
|
51
|
-
also not ``None`` or ``[]``.
|
52
|
-
experiment if ``experiment_names``
|
53
|
-
|
54
|
-
|
55
|
-
|
56
|
-
|
57
|
-
|
58
|
-
|
59
|
-
|
60
|
-
|
61
|
-
|
62
|
-
|
63
|
-
|
64
|
-
|
65
|
-
|
66
|
-
is
|
67
|
-
|
48
|
+
experiment_ids (list[str] | None): List of experiment IDs. Search can
|
49
|
+
work with experiment IDs or experiment names, but not both in the
|
50
|
+
same call. Values other than ``None`` or ``[]`` will result in
|
51
|
+
error if ``experiment_names`` is also not ``None`` or ``[]``.
|
52
|
+
``None`` will default to the active experiment if ``experiment_names``
|
53
|
+
is ``None`` or ``[]``.
|
54
|
+
experiment_ids (list[str] | None): List of experiment IDs. Search can
|
55
|
+
work with experiment IDs or experiment names, but not both in the
|
56
|
+
same call. Values other than ``None`` or ``[]`` will result in
|
57
|
+
error if ``experiment_names`` is also not ``None`` or ``[]``.
|
58
|
+
``experiment_names`` is also not ``None`` or ``[]``. ``None`` will
|
59
|
+
default to the active experiment if ``experiment_names`` is ``None``
|
60
|
+
or ``[]``.
|
61
|
+
filter_string (str): Filter query string, defaults to searching all
|
62
|
+
runs.
|
63
|
+
run_view_type (int): one of enum values ``ACTIVE_ONLY``, ``DELETED_ONLY``,
|
64
|
+
or ``ALL`` runs defined in :py:class:`mlflow.entities.ViewType`.
|
65
|
+
max_results (int): The maximum number of runs to put in the dataframe.
|
66
|
+
Default is 100,000 to avoid causing out-of-memory issues on the user's
|
67
|
+
machine.
|
68
|
+
order_by (list[str] | None): List of columns to order by (e.g.,
|
69
|
+
"metrics.rmse"). The ``order_by`` column can contain an optional
|
70
|
+
``DESC`` or ``ASC`` value. The default is ``ASC``. The default
|
71
|
+
ordering is to sort by ``start_time DESC``, then ``run_id``.
|
72
|
+
``start_time DESC``, then ``run_id``.
|
73
|
+
search_all_experiments (bool): Boolean specifying whether all
|
74
|
+
experiments should be searched. Only honored if ``experiment_ids``
|
75
|
+
is ``[]`` or ``None``.
|
76
|
+
experiment_names (list[str] | None): List of experiment names. Search
|
77
|
+
can work with experiment IDs or experiment names, but not both in
|
78
|
+
the same call. Values other than ``None`` or ``[]`` will result in
|
79
|
+
error if ``experiment_ids`` is also not ``None`` or ``[]``.
|
80
|
+
``experiment_ids`` is also not ``None`` or ``[]``. ``None`` will
|
81
|
+
default to the active experiment if ``experiment_ids`` is ``None``
|
82
|
+
or ``[]``.
|
68
83
|
|
69
84
|
Returns:
|
70
85
|
A `RunCollection` object containing the search results.
|
@@ -97,10 +112,10 @@ def list_runs(experiment_names: list[str] | None = None) -> RunCollection:
|
|
97
112
|
The returned runs are sorted by their start time in ascending order.
|
98
113
|
|
99
114
|
Args:
|
100
|
-
experiment_names: List of experiment names to search
|
101
|
-
|
102
|
-
|
103
|
-
|
115
|
+
experiment_names (list[str] | None): List of experiment names to search
|
116
|
+
for runs. If None or an empty list is provided, the function will
|
117
|
+
search the currently active experiment or all experiments except
|
118
|
+
the "Default" experiment.
|
104
119
|
|
105
120
|
Returns:
|
106
121
|
A `RunCollection` object containing the runs for the specified experiments.
|
@@ -133,27 +148,93 @@ class RunCollection:
|
|
133
148
|
def __len__(self) -> int:
|
134
149
|
return len(self._runs)
|
135
150
|
|
151
|
+
def __iter__(self) -> Iterator[Run]:
|
152
|
+
return iter(self._runs)
|
153
|
+
|
154
|
+
def __getitem__(self, index: int) -> Run:
|
155
|
+
return self._runs[index]
|
156
|
+
|
157
|
+
def __contains__(self, run: Run) -> bool:
|
158
|
+
return run in self._runs
|
159
|
+
|
160
|
+
def sort(
|
161
|
+
self,
|
162
|
+
key: Callable[[Run], Any] | None = None,
|
163
|
+
reverse: bool = False,
|
164
|
+
) -> None:
|
165
|
+
self._runs.sort(key=key or (lambda x: x.info.start_time), reverse=reverse)
|
166
|
+
|
167
|
+
def first(self) -> Run:
|
168
|
+
"""
|
169
|
+
Get the first run in the collection.
|
170
|
+
|
171
|
+
Returns:
|
172
|
+
The first run object in the collection.
|
173
|
+
|
174
|
+
Raises:
|
175
|
+
ValueError: If the collection is empty.
|
176
|
+
"""
|
177
|
+
if not self._runs:
|
178
|
+
raise ValueError("The collection is empty.")
|
179
|
+
|
180
|
+
return self._runs[0]
|
181
|
+
|
182
|
+
def try_first(self) -> Run | None:
|
183
|
+
"""
|
184
|
+
Try to get the first run in the collection.
|
185
|
+
|
186
|
+
Returns:
|
187
|
+
The first run object in the collection, or None if the collection
|
188
|
+
is empty.
|
189
|
+
"""
|
190
|
+
return self._runs[0] if self._runs else None
|
191
|
+
|
192
|
+
def last(self) -> Run:
|
193
|
+
"""
|
194
|
+
Get the last run in the collection.
|
195
|
+
|
196
|
+
Returns:
|
197
|
+
The last run object in the collection.
|
198
|
+
|
199
|
+
Raises:
|
200
|
+
ValueError: If the collection is empty.
|
201
|
+
"""
|
202
|
+
if not self._runs:
|
203
|
+
raise ValueError("The collection is empty.")
|
204
|
+
|
205
|
+
return self._runs[-1]
|
206
|
+
|
207
|
+
def try_last(self) -> Run | None:
|
208
|
+
"""
|
209
|
+
Try to get the last run in the collection.
|
210
|
+
|
211
|
+
Returns:
|
212
|
+
The last run object in the collection, or None if the collection is
|
213
|
+
empty.
|
214
|
+
"""
|
215
|
+
return self._runs[-1] if self._runs else None
|
216
|
+
|
136
217
|
def filter(self, config: object | None = None, **kwargs) -> RunCollection:
|
137
218
|
"""
|
138
219
|
Filter the runs based on the provided configuration.
|
139
220
|
|
140
221
|
This method filters the runs in the collection according to the
|
141
|
-
specified configuration object and additional key-value pairs.
|
142
|
-
|
143
|
-
|
144
|
-
|
145
|
-
|
222
|
+
specified configuration object and additional key-value pairs. The
|
223
|
+
configuration object and key-value pairs should contain key-value pairs
|
224
|
+
that correspond to the parameters of the runs. Only the runs that match
|
225
|
+
all the specified parameters will be included in the returned
|
226
|
+
`RunCollection` object.
|
146
227
|
|
147
228
|
The filtering supports:
|
148
229
|
- Exact matches for single values.
|
149
230
|
- Membership checks for lists of values.
|
150
|
-
- Range checks for tuples of two values (inclusive of the lower bound
|
151
|
-
exclusive of the upper bound).
|
231
|
+
- Range checks for tuples of two values (inclusive of the lower bound
|
232
|
+
and exclusive of the upper bound).
|
152
233
|
|
153
234
|
Args:
|
154
|
-
config: The configuration object to filter the runs.
|
155
|
-
object that provides key-value pairs through
|
156
|
-
function.
|
235
|
+
config (object | None): The configuration object to filter the runs.
|
236
|
+
This can be any object that provides key-value pairs through
|
237
|
+
the `iter_params` function.
|
157
238
|
**kwargs: Additional key-value pairs to filter the runs.
|
158
239
|
|
159
240
|
Returns:
|
@@ -161,80 +242,161 @@ class RunCollection:
|
|
161
242
|
"""
|
162
243
|
return RunCollection(filter_runs(self._runs, config, **kwargs))
|
163
244
|
|
164
|
-
def find(self, config: object | None = None, **kwargs) -> Run
|
245
|
+
def find(self, config: object | None = None, **kwargs) -> Run:
|
165
246
|
"""
|
166
247
|
Find the first run based on the provided configuration.
|
167
248
|
|
168
249
|
This method filters the runs in the collection according to the
|
169
250
|
specified configuration object and returns the first run that matches
|
170
|
-
the provided parameters. If no run matches the criteria,
|
251
|
+
the provided parameters. If no run matches the criteria, a `ValueError`
|
252
|
+
is raised.
|
171
253
|
|
172
254
|
Args:
|
173
|
-
config: The configuration object to identify the run.
|
255
|
+
config (object | None): The configuration object to identify the run.
|
174
256
|
**kwargs: Additional key-value pairs to filter the runs.
|
175
257
|
|
176
258
|
Returns:
|
177
|
-
The first run object that matches the provided configuration
|
178
|
-
|
259
|
+
The first run object that matches the provided configuration.
|
260
|
+
|
261
|
+
Raises:
|
262
|
+
ValueError: If no run matches the criteria.
|
179
263
|
|
180
264
|
See Also:
|
181
|
-
RunCollection.filter: The method that performs the actual filtering
|
265
|
+
RunCollection.filter: The method that performs the actual filtering
|
266
|
+
logic.
|
182
267
|
"""
|
183
268
|
return find_run(self._runs, config, **kwargs)
|
184
269
|
|
185
|
-
def
|
270
|
+
def try_find(self, config: object | None = None, **kwargs) -> Run | None:
|
271
|
+
"""
|
272
|
+
Find the first run based on the provided configuration.
|
273
|
+
|
274
|
+
This method filters the runs in the collection according to the
|
275
|
+
specified configuration object and returns the first run that matches
|
276
|
+
the provided parameters. If no run matches the criteria, None is
|
277
|
+
returned.
|
278
|
+
|
279
|
+
Args:
|
280
|
+
config (object | None): The configuration object to identify the run.
|
281
|
+
**kwargs: Additional key-value pairs to filter the runs.
|
282
|
+
|
283
|
+
Returns:
|
284
|
+
The first run object that matches the provided configuration, or
|
285
|
+
None if no runs match the criteria.
|
286
|
+
|
287
|
+
See Also:
|
288
|
+
RunCollection.filter: The method that performs the actual filtering
|
289
|
+
logic.
|
290
|
+
"""
|
291
|
+
return try_find_run(self._runs, config, **kwargs)
|
292
|
+
|
293
|
+
def find_last(self, config: object | None = None, **kwargs) -> Run:
|
186
294
|
"""
|
187
295
|
Find the last run based on the provided configuration.
|
188
296
|
|
189
297
|
This method filters the runs in the collection according to the
|
190
298
|
specified configuration object and returns the last run that matches
|
191
|
-
the provided parameters. If no run matches the criteria,
|
299
|
+
the provided parameters. If no run matches the criteria, a `ValueError`
|
300
|
+
is raised.
|
192
301
|
|
193
302
|
Args:
|
194
|
-
config: The configuration object to identify the run.
|
303
|
+
config (object | None): The configuration object to identify the run.
|
195
304
|
**kwargs: Additional key-value pairs to filter the runs.
|
196
305
|
|
197
306
|
Returns:
|
198
|
-
The last run object that matches the provided configuration
|
199
|
-
|
307
|
+
The last run object that matches the provided configuration.
|
308
|
+
|
309
|
+
Raises:
|
310
|
+
ValueError: If no run matches the criteria.
|
200
311
|
|
201
312
|
See Also:
|
202
|
-
RunCollection.filter: The method that performs the actual filtering
|
313
|
+
RunCollection.filter: The method that performs the actual filtering
|
314
|
+
logic.
|
203
315
|
"""
|
204
316
|
return find_last_run(self._runs, config, **kwargs)
|
205
317
|
|
206
|
-
def
|
318
|
+
def try_find_last(self, config: object | None = None, **kwargs) -> Run | None:
|
319
|
+
"""
|
320
|
+
Find the last run based on the provided configuration.
|
321
|
+
|
322
|
+
This method filters the runs in the collection according to the
|
323
|
+
specified configuration object and returns the last run that matches
|
324
|
+
the provided parameters. If no run matches the criteria, None is
|
325
|
+
returned.
|
326
|
+
|
327
|
+
Args:
|
328
|
+
config (object | None): The configuration object to identify the run.
|
329
|
+
**kwargs: Additional key-value pairs to filter the runs.
|
330
|
+
|
331
|
+
Returns:
|
332
|
+
The last run object that matches the provided configuration, or
|
333
|
+
None if no runs match the criteria.
|
334
|
+
|
335
|
+
See Also:
|
336
|
+
RunCollection.filter: The method that performs the actual filtering
|
337
|
+
logic.
|
338
|
+
"""
|
339
|
+
return try_find_last_run(self._runs, config, **kwargs)
|
340
|
+
|
341
|
+
def get(self, config: object | None = None, **kwargs) -> Run:
|
207
342
|
"""
|
208
343
|
Retrieve a specific run based on the provided configuration.
|
209
344
|
|
210
345
|
This method filters the runs in the collection according to the
|
211
|
-
specified configuration object and returns the run that matches
|
212
|
-
|
213
|
-
a `ValueError` is raised.
|
346
|
+
specified configuration object and returns the run that matches the
|
347
|
+
provided parameters. If no run matches the criteria, or if more than
|
348
|
+
one run matches the criteria, a `ValueError` is raised.
|
214
349
|
|
215
350
|
Args:
|
216
|
-
config: The configuration object to identify the run.
|
351
|
+
config (object | None): The configuration object to identify the run.
|
217
352
|
**kwargs: Additional key-value pairs to filter the runs.
|
218
353
|
|
219
354
|
Returns:
|
220
|
-
The run object that matches the provided configuration
|
221
|
-
if no runs match the criteria.
|
355
|
+
The run object that matches the provided configuration.
|
222
356
|
|
223
357
|
Raises:
|
224
|
-
ValueError: If
|
358
|
+
ValueError: If no run matches the criteria or if more than one run
|
359
|
+
matches the criteria.
|
225
360
|
|
226
361
|
See Also:
|
227
|
-
RunCollection.filter: The method that performs the actual filtering
|
362
|
+
RunCollection.filter: The method that performs the actual filtering
|
363
|
+
logic.
|
228
364
|
"""
|
229
365
|
return get_run(self._runs, config, **kwargs)
|
230
366
|
|
367
|
+
def try_get(self, config: object | None = None, **kwargs) -> Run | None:
|
368
|
+
"""
|
369
|
+
Retrieve a specific run based on the provided configuration.
|
370
|
+
|
371
|
+
This method filters the runs in the collection according to the
|
372
|
+
specified configuration object and returns the run that matches the
|
373
|
+
provided parameters. If no run matches the criteria, None is returned.
|
374
|
+
If more than one run matches the criteria, a `ValueError` is raised.
|
375
|
+
|
376
|
+
Args:
|
377
|
+
config (object | None): The configuration object to identify the run.
|
378
|
+
**kwargs: Additional key-value pairs to filter the runs.
|
379
|
+
|
380
|
+
Returns:
|
381
|
+
The run object that matches the provided configuration, or None if
|
382
|
+
no runs match the criteria.
|
383
|
+
|
384
|
+
Raises:
|
385
|
+
ValueError: If more than one run matches the criteria.
|
386
|
+
|
387
|
+
See Also:
|
388
|
+
RunCollection.filter: The method that performs the actual filtering
|
389
|
+
logic.
|
390
|
+
"""
|
391
|
+
return try_get_run(self._runs, config, **kwargs)
|
392
|
+
|
231
393
|
def get_param_names(self) -> list[str]:
|
232
394
|
"""
|
233
395
|
Get the parameter names from the runs.
|
234
396
|
|
235
|
-
This method extracts the unique parameter names from the provided list
|
236
|
-
It iterates through each run and collects the parameter names
|
237
|
-
ensure uniqueness.
|
397
|
+
This method extracts the unique parameter names from the provided list
|
398
|
+
of runs. It iterates through each run and collects the parameter names
|
399
|
+
into a set to ensure uniqueness.
|
238
400
|
|
239
401
|
Returns:
|
240
402
|
A list of unique parameter names.
|
@@ -246,101 +408,95 @@ class RunCollection:
|
|
246
408
|
Get the parameter dictionary from the list of runs.
|
247
409
|
|
248
410
|
This method extracts the parameter names and their corresponding values
|
249
|
-
from the provided list of runs. It iterates through each run and
|
250
|
-
the parameter values into a dictionary where the keys are
|
251
|
-
and the values are lists of parameter values.
|
411
|
+
from the provided list of runs. It iterates through each run and
|
412
|
+
collects the parameter values into a dictionary where the keys are
|
413
|
+
parameter names and the values are lists of parameter values.
|
252
414
|
|
253
415
|
Returns:
|
254
|
-
A dictionary where the keys are parameter names and the values are
|
255
|
-
of parameter values.
|
416
|
+
A dictionary where the keys are parameter names and the values are
|
417
|
+
lists of parameter values.
|
256
418
|
"""
|
257
419
|
return get_param_dict(self._runs)
|
258
420
|
|
259
|
-
def first(self) -> Run | None:
|
260
|
-
"""
|
261
|
-
Return the first run in the collection.
|
262
|
-
|
263
|
-
Returns:
|
264
|
-
The first Run object if the collection is not empty, otherwise None.
|
265
|
-
"""
|
266
|
-
return self._runs[0] if self._runs else None
|
267
|
-
|
268
|
-
def last(self) -> Run | None:
|
269
|
-
"""
|
270
|
-
Return the last run in the collection.
|
271
|
-
|
272
|
-
Returns:
|
273
|
-
The last Run object if the collection is not empty, otherwise None.
|
274
|
-
"""
|
275
|
-
return self._runs[-1] if self._runs else None
|
276
|
-
|
277
421
|
def map(self, func: Callable[[Run], T]) -> Iterator[T]:
|
278
422
|
"""
|
279
|
-
Apply a function to each run in the collection and return an iterator of
|
423
|
+
Apply a function to each run in the collection and return an iterator of
|
424
|
+
results.
|
280
425
|
|
281
426
|
Args:
|
282
|
-
func: A function that takes a
|
427
|
+
func (Callable[[Run], T]): A function that takes a run and returns a
|
428
|
+
result.
|
283
429
|
|
284
|
-
|
285
|
-
|
286
|
-
|
430
|
+
Yields:
|
431
|
+
Results obtained by applying the function to each run in the
|
432
|
+
collection.
|
287
433
|
"""
|
288
434
|
return (func(run) for run in self._runs)
|
289
435
|
|
290
436
|
def map_run_id(self, func: Callable[[str], T]) -> Iterator[T]:
|
291
437
|
"""
|
292
|
-
Apply a function to each run id in the collection and return an iterator
|
438
|
+
Apply a function to each run id in the collection and return an iterator
|
439
|
+
of results.
|
293
440
|
|
294
441
|
Args:
|
295
|
-
func: A function that takes a run id and returns a
|
442
|
+
func (Callable[[str], T]): A function that takes a run id and returns a
|
443
|
+
result.
|
296
444
|
|
297
|
-
|
298
|
-
|
299
|
-
|
445
|
+
Yields:
|
446
|
+
Results obtained by applying the function to each run id in the
|
447
|
+
collection.
|
300
448
|
"""
|
301
449
|
return (func(run.info.run_id) for run in self._runs)
|
302
450
|
|
303
451
|
def map_config(self, func: Callable[[DictConfig], T]) -> Iterator[T]:
|
304
452
|
"""
|
305
|
-
Apply a function to each run
|
453
|
+
Apply a function to each run configuration in the collection and return
|
454
|
+
an iterator of results.
|
306
455
|
|
307
456
|
Args:
|
308
|
-
func: A function that takes a run
|
457
|
+
func (Callable[[DictConfig], T]): A function that takes a run
|
458
|
+
configuration and returns a result.
|
309
459
|
|
310
|
-
|
311
|
-
|
460
|
+
Yields:
|
461
|
+
Results obtained by applying the function to each run configuration
|
312
462
|
in the collection.
|
313
463
|
"""
|
314
464
|
return (func(load_config(run)) for run in self._runs)
|
315
465
|
|
316
466
|
def map_uri(self, func: Callable[[str | None], T]) -> Iterator[T]:
|
317
467
|
"""
|
318
|
-
Apply a function to each artifact URI in the collection and return an
|
468
|
+
Apply a function to each artifact URI in the collection and return an
|
469
|
+
iterator of results.
|
319
470
|
|
320
|
-
This method iterates over each run in the collection, retrieves the
|
321
|
-
and applies the provided function to it. If a run does not
|
322
|
-
None is passed to the function.
|
471
|
+
This method iterates over each run in the collection, retrieves the
|
472
|
+
artifact URI, and applies the provided function to it. If a run does not
|
473
|
+
have an artifact URI, None is passed to the function.
|
323
474
|
|
324
475
|
Args:
|
325
|
-
func: A function that takes an
|
476
|
+
func (Callable[[str | None], T]): A function that takes an
|
477
|
+
artifact URI (string or None) and returns a result.
|
326
478
|
|
327
479
|
Yields:
|
328
|
-
|
480
|
+
Results obtained by applying the function to each artifact URI in the
|
481
|
+
collection.
|
329
482
|
"""
|
330
483
|
return (func(run.info.artifact_uri) for run in self._runs)
|
331
484
|
|
332
485
|
def map_dir(self, func: Callable[[str], T]) -> Iterator[T]:
|
333
486
|
"""
|
334
|
-
Apply a function to each artifact directory in the collection and return
|
487
|
+
Apply a function to each artifact directory in the collection and return
|
488
|
+
an iterator of results.
|
335
489
|
|
336
|
-
This method iterates over each run in the collection, downloads the
|
337
|
-
and applies the provided function to the directory
|
490
|
+
This method iterates over each run in the collection, downloads the
|
491
|
+
artifact directory, and applies the provided function to the directory
|
492
|
+
path.
|
338
493
|
|
339
494
|
Args:
|
340
|
-
func: A function that takes an artifact directory
|
495
|
+
func (Callable[[str], T]): A function that takes an artifact directory
|
496
|
+
path (string) and returns a result.
|
341
497
|
|
342
|
-
|
343
|
-
|
498
|
+
Yields:
|
499
|
+
Results obtained by applying the function to each artifact directory
|
344
500
|
in the collection.
|
345
501
|
"""
|
346
502
|
return (func(download_artifacts(run_id=run.info.run_id)) for run in self._runs)
|
@@ -350,25 +506,21 @@ def _param_matches(run: Run, key: str, value: Any) -> bool:
|
|
350
506
|
"""
|
351
507
|
Check if the run's parameter matches the specified key-value pair.
|
352
508
|
|
353
|
-
This function checks if the run's parameters contain the specified
|
354
|
-
It handles different types of values, including lists
|
509
|
+
This function checks if the run's parameters contain the specified
|
510
|
+
key-value pair. It handles different types of values, including lists
|
511
|
+
and tuples.
|
355
512
|
|
356
513
|
Args:
|
357
|
-
run: The run object to check.
|
358
|
-
key: The parameter key to check.
|
359
|
-
value: The parameter value to check.
|
514
|
+
run (Run): The run object to check.
|
515
|
+
key (str): The parameter key to check.
|
516
|
+
value (Any): The parameter value to check.
|
360
517
|
|
361
518
|
Returns:
|
362
|
-
True if the run's parameter matches the specified key-value pair,
|
519
|
+
True if the run's parameter matches the specified key-value pair,
|
520
|
+
False otherwise.
|
363
521
|
"""
|
364
522
|
param = run.data.params.get(key, value)
|
365
523
|
|
366
|
-
# FIXME: This is a workaround to handle the case where the parameter value is a list
|
367
|
-
# We need to improve the logic to handle different types of values
|
368
|
-
# For now, we assume that if the parameter is a list, we should check if it contains the value
|
369
|
-
# This is not ideal, but it works for the case where the parameter value is a list of strings
|
370
|
-
# We should improve the logic to handle different types of values in the future
|
371
|
-
|
372
524
|
if param is None:
|
373
525
|
return False
|
374
526
|
|
@@ -392,8 +544,8 @@ def filter_runs(runs: list[Run], config: object | None = None, **kwargs) -> list
|
|
392
544
|
specified configuration object and additional key-value pairs.
|
393
545
|
The configuration object and key-value pairs should contain
|
394
546
|
key-value pairs that correspond to the parameters of the runs.
|
395
|
-
Only the runs that match all the specified parameters will
|
396
|
-
included in the returned list of runs.
|
547
|
+
Only the runs that match all the specified parameters will
|
548
|
+
be included in the returned list of runs.
|
397
549
|
|
398
550
|
The filtering supports:
|
399
551
|
- Exact matches for single values.
|
@@ -402,9 +554,10 @@ def filter_runs(runs: list[Run], config: object | None = None, **kwargs) -> list
|
|
402
554
|
exclusive of the upper bound).
|
403
555
|
|
404
556
|
Args:
|
405
|
-
runs: The list of runs to filter.
|
406
|
-
config: The configuration object to filter the runs.
|
407
|
-
|
557
|
+
runs (list[Run]): The list of runs to filter.
|
558
|
+
config (object | None): The configuration object to filter the runs.
|
559
|
+
This can be any object that provides key-value pairs through the
|
560
|
+
`iter_params` function.
|
408
561
|
**kwargs: Additional key-value pairs to filter the runs.
|
409
562
|
|
410
563
|
Returns:
|
@@ -419,7 +572,38 @@ def filter_runs(runs: list[Run], config: object | None = None, **kwargs) -> list
|
|
419
572
|
return runs
|
420
573
|
|
421
574
|
|
422
|
-
def find_run(runs: list[Run], config: object | None = None, **kwargs) -> Run
|
575
|
+
def find_run(runs: list[Run], config: object | None = None, **kwargs) -> Run:
|
576
|
+
"""
|
577
|
+
Find the first run based on the provided configuration.
|
578
|
+
|
579
|
+
This method filters the runs in the collection according to the
|
580
|
+
specified configuration object and returns the first run that matches
|
581
|
+
the provided parameters. If no run matches the criteria, a `ValueError` is
|
582
|
+
raised.
|
583
|
+
|
584
|
+
Args:
|
585
|
+
runs (list[Run]): The runs to filter.
|
586
|
+
config (object | None): The configuration object to identify the run.
|
587
|
+
**kwargs: Additional key-value pairs to filter the runs.
|
588
|
+
|
589
|
+
Returns:
|
590
|
+
The first run object that matches the provided configuration.
|
591
|
+
|
592
|
+
Raises:
|
593
|
+
ValueError: If no run matches the criteria.
|
594
|
+
|
595
|
+
See Also:
|
596
|
+
RunCollection.filter: The method that performs the actual filtering logic.
|
597
|
+
"""
|
598
|
+
filtered_runs = filter_runs(runs, config, **kwargs)
|
599
|
+
|
600
|
+
if len(filtered_runs) == 0:
|
601
|
+
raise ValueError("No run matches the provided configuration.")
|
602
|
+
|
603
|
+
return filtered_runs[0]
|
604
|
+
|
605
|
+
|
606
|
+
def try_find_run(runs: list[Run], config: object | None = None, **kwargs) -> Run | None:
|
423
607
|
"""
|
424
608
|
Find the first run based on the provided configuration.
|
425
609
|
|
@@ -428,19 +612,55 @@ def find_run(runs: list[Run], config: object | None = None, **kwargs) -> Run | N
|
|
428
612
|
the provided parameters. If no run matches the criteria, None is returned.
|
429
613
|
|
430
614
|
Args:
|
431
|
-
runs: The runs to filter.
|
432
|
-
config: The configuration object to identify the run.
|
615
|
+
runs (list[Run]): The runs to filter.
|
616
|
+
config (object | None): The configuration object to identify the run.
|
433
617
|
**kwargs: Additional key-value pairs to filter the runs.
|
434
618
|
|
435
619
|
Returns:
|
436
620
|
The first run object that matches the provided configuration, or None
|
437
621
|
if no runs match the criteria.
|
438
622
|
"""
|
439
|
-
|
440
|
-
|
623
|
+
filtered_runs = filter_runs(runs, config, **kwargs)
|
624
|
+
|
625
|
+
if len(filtered_runs) == 0:
|
626
|
+
return None
|
627
|
+
|
628
|
+
return filtered_runs[0]
|
629
|
+
|
630
|
+
|
631
|
+
def find_last_run(runs: list[Run], config: object | None = None, **kwargs) -> Run:
|
632
|
+
"""
|
633
|
+
Find the last run based on the provided configuration.
|
634
|
+
|
635
|
+
This method filters the runs in the collection according to the
|
636
|
+
specified configuration object and returns the last run that matches
|
637
|
+
the provided parameters. If no run matches the criteria, a `ValueError`
|
638
|
+
is raised.
|
639
|
+
|
640
|
+
Args:
|
641
|
+
runs (list[Run]): The runs to filter.
|
642
|
+
config (object | None): The configuration object to identify the run.
|
643
|
+
**kwargs: Additional key-value pairs to filter the runs.
|
644
|
+
|
645
|
+
Returns:
|
646
|
+
The last run object that matches the provided configuration.
|
647
|
+
|
648
|
+
Raises:
|
649
|
+
ValueError: If no run matches the criteria.
|
650
|
+
|
651
|
+
See Also:
|
652
|
+
RunCollection.filter: The method that performs the actual filtering
|
653
|
+
logic.
|
654
|
+
"""
|
655
|
+
filtered_runs = filter_runs(runs, config, **kwargs)
|
441
656
|
|
657
|
+
if len(filtered_runs) == 0:
|
658
|
+
raise ValueError("No run matches the provided configuration.")
|
442
659
|
|
443
|
-
|
660
|
+
return filtered_runs[-1]
|
661
|
+
|
662
|
+
|
663
|
+
def try_find_last_run(runs: list[Run], config: object | None = None, **kwargs) -> Run | None:
|
444
664
|
"""
|
445
665
|
Find the last run based on the provided configuration.
|
446
666
|
|
@@ -449,30 +669,74 @@ def find_last_run(runs: list[Run], config: object | None = None, **kwargs) -> Ru
|
|
449
669
|
the provided parameters. If no run matches the criteria, None is returned.
|
450
670
|
|
451
671
|
Args:
|
452
|
-
runs: The runs to filter.
|
453
|
-
config: The configuration object to identify the run.
|
672
|
+
runs (list[Run]): The runs to filter.
|
673
|
+
config (object | None): The configuration object to identify the run.
|
454
674
|
**kwargs: Additional key-value pairs to filter the runs.
|
455
675
|
|
456
676
|
Returns:
|
457
677
|
The last run object that matches the provided configuration, or None
|
458
678
|
if no runs match the criteria.
|
459
679
|
"""
|
460
|
-
|
461
|
-
|
680
|
+
filtered_runs = filter_runs(runs, config, **kwargs)
|
681
|
+
|
682
|
+
if len(filtered_runs) == 0:
|
683
|
+
return None
|
684
|
+
|
685
|
+
return filtered_runs[-1]
|
462
686
|
|
463
687
|
|
464
|
-
def get_run(runs: list[Run], config: object | None = None, **kwargs) -> Run
|
688
|
+
def get_run(runs: list[Run], config: object | None = None, **kwargs) -> Run:
|
465
689
|
"""
|
466
690
|
Retrieve a specific run based on the provided configuration.
|
467
691
|
|
468
692
|
This method filters the runs in the collection according to the
|
469
693
|
specified configuration object and returns the run that matches
|
470
|
-
the provided parameters. If
|
471
|
-
a `ValueError` is raised.
|
694
|
+
the provided parameters. If no run matches the criteria, or if more
|
695
|
+
than one run matches the criteria, a `ValueError` is raised.
|
472
696
|
|
473
697
|
Args:
|
474
|
-
runs: The runs to filter.
|
475
|
-
config: The configuration object to identify the run.
|
698
|
+
runs (list[Run]): The runs to filter.
|
699
|
+
config (object | None): The configuration object to identify the run.
|
700
|
+
**kwargs: Additional key-value pairs to filter the runs.
|
701
|
+
|
702
|
+
Returns:
|
703
|
+
The run object that matches the provided configuration.
|
704
|
+
|
705
|
+
Raises:
|
706
|
+
ValueError: If no run matches the criteria or if more than one run
|
707
|
+
matches the criteria.
|
708
|
+
|
709
|
+
See Also:
|
710
|
+
RunCollection.filter: The method that performs the actual filtering
|
711
|
+
logic.
|
712
|
+
"""
|
713
|
+
filtered_runs = filter_runs(runs, config, **kwargs)
|
714
|
+
|
715
|
+
if len(filtered_runs) == 0:
|
716
|
+
raise ValueError("No run matches the provided configuration.")
|
717
|
+
|
718
|
+
if len(filtered_runs) == 1:
|
719
|
+
return filtered_runs[0]
|
720
|
+
|
721
|
+
msg = (
|
722
|
+
f"Multiple runs were filtered. Expected number of runs is 1, "
|
723
|
+
f"but found {len(filtered_runs)} runs."
|
724
|
+
)
|
725
|
+
raise ValueError(msg)
|
726
|
+
|
727
|
+
|
728
|
+
def try_get_run(runs: list[Run], config: object | None = None, **kwargs) -> Run | None:
|
729
|
+
"""
|
730
|
+
Retrieve a specific run based on the provided configuration.
|
731
|
+
|
732
|
+
This method filters the runs in the collection according to the
|
733
|
+
specified configuration object and returns the run that matches
|
734
|
+
the provided parameters. If no run matches the criteria, None is returned.
|
735
|
+
If more than one run matches the criteria, a `ValueError` is raised.
|
736
|
+
|
737
|
+
Args:
|
738
|
+
runs (list[Run]): The runs to filter.
|
739
|
+
config (object | None): The configuration object to identify the run.
|
476
740
|
**kwargs: Additional key-value pairs to filter the runs.
|
477
741
|
|
478
742
|
Returns:
|
@@ -481,16 +745,23 @@ def get_run(runs: list[Run], config: object | None = None, **kwargs) -> Run | No
|
|
481
745
|
|
482
746
|
Raises:
|
483
747
|
ValueError: If more than one run matches the criteria.
|
748
|
+
|
749
|
+
See Also:
|
750
|
+
RunCollection.filter: The method that performs the actual filtering
|
751
|
+
logic.
|
484
752
|
"""
|
485
|
-
|
753
|
+
filtered_runs = filter_runs(runs, config, **kwargs)
|
486
754
|
|
487
|
-
if len(
|
755
|
+
if len(filtered_runs) == 0:
|
488
756
|
return None
|
489
757
|
|
490
|
-
if len(
|
491
|
-
return
|
758
|
+
if len(filtered_runs) == 1:
|
759
|
+
return filtered_runs[0]
|
492
760
|
|
493
|
-
msg =
|
761
|
+
msg = (
|
762
|
+
"Multiple runs were filtered. Expected number of runs is 1, "
|
763
|
+
f"but found {len(filtered_runs)} runs."
|
764
|
+
)
|
494
765
|
raise ValueError(msg)
|
495
766
|
|
496
767
|
|
@@ -498,12 +769,12 @@ def get_param_names(runs: list[Run]) -> list[str]:
|
|
498
769
|
"""
|
499
770
|
Get the parameter names from the runs.
|
500
771
|
|
501
|
-
This method extracts the unique parameter names from the provided list of
|
502
|
-
It iterates through each run and collects the parameter names into a
|
503
|
-
ensure uniqueness.
|
772
|
+
This method extracts the unique parameter names from the provided list of
|
773
|
+
runs. It iterates through each run and collects the parameter names into a
|
774
|
+
set to ensure uniqueness.
|
504
775
|
|
505
776
|
Args:
|
506
|
-
runs: The list of runs from which to extract parameter names.
|
777
|
+
runs (list[Run]): The list of runs from which to extract parameter names.
|
507
778
|
|
508
779
|
Returns:
|
509
780
|
A list of unique parameter names.
|
@@ -527,7 +798,8 @@ def get_param_dict(runs: list[Run]) -> dict[str, list[str]]:
|
|
527
798
|
and the values are lists of parameter values.
|
528
799
|
|
529
800
|
Args:
|
530
|
-
runs: The list of runs from which to extract parameter names
|
801
|
+
runs (list[Run]): The list of runs from which to extract parameter names
|
802
|
+
and values.
|
531
803
|
|
532
804
|
Returns:
|
533
805
|
A dictionary where the keys are parameter names and the values are lists
|
@@ -552,7 +824,7 @@ def load_config(run: Run) -> DictConfig:
|
|
552
824
|
`.hydra/config.yaml` is not found in the run's artifact directory.
|
553
825
|
|
554
826
|
Args:
|
555
|
-
run: The Run instance for which to load the configuration.
|
827
|
+
run (Run): The Run instance for which to load the configuration.
|
556
828
|
|
557
829
|
Returns:
|
558
830
|
The loaded configuration as a DictConfig object. Returns an empty
|
@@ -570,37 +842,3 @@ def _load_config(run_id: str) -> DictConfig:
|
|
570
842
|
return DictConfig({})
|
571
843
|
|
572
844
|
return OmegaConf.load(path) # type: ignore
|
573
|
-
|
574
|
-
|
575
|
-
# def get_hydra_output_dir(run: Run_ | Series | str) -> Path:
|
576
|
-
# """
|
577
|
-
# Get the Hydra output directory.
|
578
|
-
|
579
|
-
# Args:
|
580
|
-
# run: The run object.
|
581
|
-
|
582
|
-
# Returns:
|
583
|
-
# Path: The Hydra output directory.
|
584
|
-
# """
|
585
|
-
# path = get_artifact_dir(run) / ".hydra/hydra.yaml"
|
586
|
-
|
587
|
-
# if path.exists():
|
588
|
-
# hc = OmegaConf.load(path)
|
589
|
-
# return Path(hc.hydra.runtime.output_dir)
|
590
|
-
|
591
|
-
# raise FileNotFoundError
|
592
|
-
|
593
|
-
|
594
|
-
# def log_hydra_output_dir(run: Run_ | Series | str) -> None:
|
595
|
-
# """
|
596
|
-
# Log the Hydra output directory.
|
597
|
-
|
598
|
-
# Args:
|
599
|
-
# run: The run object.
|
600
|
-
|
601
|
-
# Returns:
|
602
|
-
# None
|
603
|
-
# """
|
604
|
-
# output_dir = get_hydra_output_dir(run)
|
605
|
-
# run_id = run if isinstance(run, str) else run.info.run_id
|
606
|
-
# mlflow.log_artifacts(output_dir.as_posix(), run_id=run_id)
|